From 4eea22f01bb4fdba1aab4430c33adbe88d9d4985 Mon Sep 17 00:00:00 2001 From: Mandy Kirkconnell Date: Tue, 14 Mar 2006 13:29:52 +1100 Subject: [XFS] 929045 567344 This mod re-organizes some of the in-core file extent code to prepare for an upcoming mod which will introduce multi-level in-core extent allocations. Although the in-core extent management is using a new code path in this mod, the functionality remains the same. Major changes include: - Introduce 10 new subroutines which re-orgainze the existing code but do NOT change functionality: xfs_iext_get_ext() xfs_iext_insert() xfs_iext_add() xfs_iext_remove() xfs_iext_remove_inline() xfs_iext_remove_direct() xfs_iext_realloc_direct() xfs_iext_direct_to_inline() xfs_iext_inline_to_direct() xfs_iext_destroy() - Remove 2 subroutines (functionality moved to new subroutines above): xfs_iext_realloc() -replaced by xfs_iext_add() and xfs_iext_remove() xfs_bmap_insert_exlist() - replaced by xfs_iext_insert() xfs_bmap_delete_exlist() - replaced by xfs_iext_remove() - Replace all hard-coded (indexed) extent assignments with a call to xfs_iext_get_ext() - Replace all extent record pointer arithmetic (ep++, ep--, base + lastx,..) with calls to xfs_iext_get_ext() - Update comments to remove the idea of a single "extent list" and introduce "extent record" terminology instead SGI-PV: 928864 SGI-Modid: xfs-linux-melb:xfs-kern:207390a Signed-off-by: Mandy Kirkconnell Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 496 ++++++++++++++++++++++++------------------------------ 1 file changed, 219 insertions(+), 277 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index 70625e577c7..53c47a181f8 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -89,7 +89,7 @@ xfs_bmap_add_attrfork_local( int *flags); /* inode logging flags */ /* - * Called by xfs_bmapi to update extent list structure and the btree + * Called by xfs_bmapi to update file extent records and the btree * after allocating space (or doing a delayed allocation). */ STATIC int /* error */ @@ -97,7 +97,7 @@ xfs_bmap_add_extent( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ xfs_fsblock_t *first, /* pointer to firstblock variable */ xfs_bmap_free_t *flist, /* list of extents to be freed */ int *logflagsp, /* inode logging flags */ @@ -113,7 +113,7 @@ xfs_bmap_add_extent_delay_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ xfs_filblks_t *dnew, /* new delayed-alloc indirect blocks */ xfs_fsblock_t *first, /* pointer to firstblock variable */ xfs_bmap_free_t *flist, /* list of extents to be freed */ @@ -129,7 +129,7 @@ xfs_bmap_add_extent_hole_delay( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp,/* inode logging flags */ int rsvd); /* OK to allocate reserved blocks */ @@ -142,7 +142,7 @@ xfs_bmap_add_extent_hole_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp, /* inode logging flags */ int whichfork); /* data or attr fork */ @@ -155,7 +155,7 @@ xfs_bmap_add_extent_unwritten_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp); /* inode logging flags */ /* @@ -169,7 +169,7 @@ xfs_bmap_alloc( /* * Transform a btree format file with only one leaf node, where the * extents list will fit in the inode, into an extents format file. - * Since the extent list is already in-core, all we have to do is + * Since the file extents are already in-core, all we have to do is * give up the space for the btree root and pitch the leaf block. */ STATIC int /* error */ @@ -191,7 +191,7 @@ xfs_bmap_check_extents( #endif /* - * Called by xfs_bmapi to update extent list structure and the btree + * Called by xfs_bmapi to update file extent records and the btree * after removing space (or undoing a delayed allocation). */ STATIC int /* error */ @@ -201,7 +201,7 @@ xfs_bmap_del_extent( xfs_extnum_t idx, /* extent number to update/insert */ xfs_bmap_free_t *flist, /* list of extents to be freed */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp,/* inode logging flags */ int whichfork, /* data or attr fork */ int rsvd); /* OK to allocate reserved blocks */ @@ -216,18 +216,6 @@ xfs_bmap_del_free( xfs_bmap_free_item_t *prev, /* previous item on list, if any */ xfs_bmap_free_item_t *free); /* list item to be freed */ -/* - * Remove count entries from the extents array for inode "ip", starting - * at index "idx". Copies the remaining items down over the deleted ones, - * and gives back the excess memory. - */ -STATIC void -xfs_bmap_delete_exlist( - xfs_inode_t *ip, /* incode inode pointer */ - xfs_extnum_t idx, /* starting delete index */ - xfs_extnum_t count, /* count of items to delete */ - int whichfork); /* data or attr fork */ - /* * Convert an extents-format file into a btree-format file. * The new file will have a root block (in the inode) and a single child block. @@ -243,18 +231,6 @@ xfs_bmap_extents_to_btree( int *logflagsp, /* inode logging flags */ int whichfork); /* data or attr fork */ -/* - * Insert new item(s) in the extent list for inode "ip". - * Count new items are inserted at offset idx. - */ -STATIC void -xfs_bmap_insert_exlist( - xfs_inode_t *ip, /* incore inode pointer */ - xfs_extnum_t idx, /* starting index of new items */ - xfs_extnum_t count, /* number of inserted items */ - xfs_bmbt_irec_t *new, /* items to insert */ - int whichfork); /* data or attr fork */ - /* * Convert a local file to an extents file. * This code is sort of bogus, since the file data needs to get @@ -316,7 +292,7 @@ xfs_bmap_trace_addentry( int whichfork); /* data or attr fork */ /* - * Add bmap trace entry prior to a call to xfs_bmap_delete_exlist. + * Add bmap trace entry prior to a call to xfs_iext_remove. */ STATIC void xfs_bmap_trace_delete( @@ -328,7 +304,7 @@ xfs_bmap_trace_delete( int whichfork); /* data or attr fork */ /* - * Add bmap trace entry prior to a call to xfs_bmap_insert_exlist, or + * Add bmap trace entry prior to a call to xfs_iext_insert, or * reading in the extents list from the disk (in the btree). */ STATIC void @@ -343,7 +319,7 @@ xfs_bmap_trace_insert( int whichfork); /* data or attr fork */ /* - * Add bmap trace entry after updating an extent list entry in place. + * Add bmap trace entry after updating an extent record in place. */ STATIC void xfs_bmap_trace_post_update( @@ -354,7 +330,7 @@ xfs_bmap_trace_post_update( int whichfork); /* data or attr fork */ /* - * Add bmap trace entry prior to updating an extent list entry in place. + * Add bmap trace entry prior to updating an extent record in place. */ STATIC void xfs_bmap_trace_pre_update( @@ -413,19 +389,24 @@ STATIC int xfs_bmap_count_tree( xfs_mount_t *mp, xfs_trans_t *tp, + xfs_ifork_t *ifp, xfs_fsblock_t blockno, int levelin, int *count); STATIC int xfs_bmap_count_leaves( - xfs_bmbt_rec_t *frp, + xfs_ifork_t *ifp, + xfs_extnum_t idx, int numrecs, int *count); STATIC int xfs_bmap_disk_count_leaves( - xfs_bmbt_rec_t *frp, + xfs_ifork_t *ifp, + xfs_mount_t *mp, + xfs_extnum_t idx, + xfs_bmbt_block_t *block, int numrecs, int *count); @@ -537,7 +518,7 @@ xfs_bmap_add_attrfork_local( } /* - * Called by xfs_bmapi to update extent list structure and the btree + * Called by xfs_bmapi to update file extent records and the btree * after allocating space (or doing a delayed allocation). */ STATIC int /* error */ @@ -545,7 +526,7 @@ xfs_bmap_add_extent( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ xfs_fsblock_t *first, /* pointer to firstblock variable */ xfs_bmap_free_t *flist, /* list of extents to be freed */ int *logflagsp, /* inode logging flags */ @@ -578,7 +559,7 @@ xfs_bmap_add_extent( if (nextents == 0) { xfs_bmap_trace_insert(fname, "insert empty", ip, 0, 1, new, NULL, whichfork); - xfs_bmap_insert_exlist(ip, 0, 1, new, whichfork); + xfs_iext_insert(ifp, 0, 1, new); ASSERT(cur == NULL); ifp->if_lastex = 0; if (!ISNULLSTARTBLOCK(new->br_startblock)) { @@ -614,7 +595,7 @@ xfs_bmap_add_extent( /* * Get the record referred to by idx. */ - xfs_bmbt_get_all(&ifp->if_u1.if_extents[idx], &prev); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx), &prev); /* * If it's a real allocation record, and the new allocation ends * after the start of the referred to record, then we're filling @@ -714,14 +695,13 @@ xfs_bmap_add_extent_delay_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ xfs_filblks_t *dnew, /* new delayed-alloc indirect blocks */ xfs_fsblock_t *first, /* pointer to firstblock variable */ xfs_bmap_free_t *flist, /* list of extents to be freed */ int *logflagsp, /* inode logging flags */ int rsvd) /* OK to use reserved data block allocation */ { - xfs_bmbt_rec_t *base; /* base of extent entry list */ xfs_btree_cur_t *cur; /* btree cursor */ int diff; /* temp value */ xfs_bmbt_rec_t *ep; /* extent entry for idx */ @@ -730,6 +710,7 @@ xfs_bmap_add_extent_delay_real( static char fname[] = "xfs_bmap_add_extent_delay_real"; #endif int i; /* temp state */ + xfs_ifork_t *ifp; /* inode fork pointer */ xfs_fileoff_t new_endoff; /* end offset of new entry */ xfs_bmbt_irec_t r[3]; /* neighbor extent entries */ /* left is 0, right is 1, prev is 2 */ @@ -763,8 +744,8 @@ xfs_bmap_add_extent_delay_real( * Set up a bunch of variables to make the tests simpler. */ cur = *curp; - base = ip->i_df.if_u1.if_extents; - ep = &base[idx]; + ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK); + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_get_all(ep, &PREV); new_endoff = new->br_startoff + new->br_blockcount; ASSERT(PREV.br_startoff <= new->br_startoff); @@ -781,7 +762,7 @@ xfs_bmap_add_extent_delay_real( * Don't set contiguous if the combined extent would be too large. */ if (STATE_SET_TEST(LEFT_VALID, idx > 0)) { - xfs_bmbt_get_all(ep - 1, &LEFT); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx - 1), &LEFT); STATE_SET(LEFT_DELAY, ISNULLSTARTBLOCK(LEFT.br_startblock)); } STATE_SET(LEFT_CONTIG, @@ -798,7 +779,7 @@ xfs_bmap_add_extent_delay_real( if (STATE_SET_TEST(RIGHT_VALID, idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1)) { - xfs_bmbt_get_all(ep + 1, &RIGHT); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx + 1), &RIGHT); STATE_SET(RIGHT_DELAY, ISNULLSTARTBLOCK(RIGHT.br_startblock)); } STATE_SET(RIGHT_CONTIG, @@ -825,14 +806,14 @@ xfs_bmap_add_extent_delay_real( */ xfs_bmap_trace_pre_update(fname, "LF|RF|LC|RC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + PREV.br_blockcount + RIGHT.br_blockcount); xfs_bmap_trace_post_update(fname, "LF|RF|LC|RC", ip, idx - 1, XFS_DATA_FORK); xfs_bmap_trace_delete(fname, "LF|RF|LC|RC", ip, idx, 2, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx, 2, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx, 2); ip->i_df.if_lastex = idx - 1; ip->i_d.di_nextents--; if (cur == NULL) @@ -867,14 +848,14 @@ xfs_bmap_add_extent_delay_real( */ xfs_bmap_trace_pre_update(fname, "LF|RF|LC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + PREV.br_blockcount); xfs_bmap_trace_post_update(fname, "LF|RF|LC", ip, idx - 1, XFS_DATA_FORK); ip->i_df.if_lastex = idx - 1; xfs_bmap_trace_delete(fname, "LF|RF|LC", ip, idx, 1, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx, 1, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx, 1); if (cur == NULL) rval = XFS_ILOG_DEXT; else { @@ -908,7 +889,7 @@ xfs_bmap_add_extent_delay_real( ip->i_df.if_lastex = idx; xfs_bmap_trace_delete(fname, "LF|RF|RC", ip, idx + 1, 1, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx + 1, 1, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx + 1, 1); if (cur == NULL) rval = XFS_ILOG_DEXT; else { @@ -964,7 +945,7 @@ xfs_bmap_add_extent_delay_real( */ xfs_bmap_trace_pre_update(fname, "LF|LC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + new->br_blockcount); xfs_bmbt_set_startoff(ep, PREV.br_startoff + new->br_blockcount); @@ -1010,7 +991,7 @@ xfs_bmap_add_extent_delay_real( xfs_bmbt_set_blockcount(ep, temp); xfs_bmap_trace_insert(fname, "LF", ip, idx, 1, new, NULL, XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx, 1, new, XFS_DATA_FORK); + xfs_iext_insert(ifp, idx, 1, new); ip->i_df.if_lastex = idx; ip->i_d.di_nextents++; if (cur == NULL) @@ -1039,8 +1020,7 @@ xfs_bmap_add_extent_delay_real( temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp), STARTBLOCKVAL(PREV.br_startblock) - (cur ? cur->bc_private.b.allocated : 0)); - base = ip->i_df.if_u1.if_extents; - ep = &base[idx + 1]; + ep = xfs_iext_get_ext(ifp, idx + 1); xfs_bmbt_set_startblock(ep, NULLSTARTBLOCK((int)temp)); xfs_bmap_trace_post_update(fname, "LF", ip, idx + 1, XFS_DATA_FORK); @@ -1058,7 +1038,8 @@ xfs_bmap_add_extent_delay_real( xfs_bmap_trace_pre_update(fname, "RF|RC", ip, idx + 1, XFS_DATA_FORK); xfs_bmbt_set_blockcount(ep, temp); - xfs_bmbt_set_allf(ep + 1, new->br_startoff, new->br_startblock, + xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, idx + 1), + new->br_startoff, new->br_startblock, new->br_blockcount + RIGHT.br_blockcount, RIGHT.br_state); xfs_bmap_trace_post_update(fname, "RF|RC", ip, idx + 1, @@ -1098,7 +1079,7 @@ xfs_bmap_add_extent_delay_real( xfs_bmbt_set_blockcount(ep, temp); xfs_bmap_trace_insert(fname, "RF", ip, idx + 1, 1, new, NULL, XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx + 1, 1, new, XFS_DATA_FORK); + xfs_iext_insert(ifp, idx + 1, 1, new); ip->i_df.if_lastex = idx + 1; ip->i_d.di_nextents++; if (cur == NULL) @@ -1127,8 +1108,7 @@ xfs_bmap_add_extent_delay_real( temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp), STARTBLOCKVAL(PREV.br_startblock) - (cur ? cur->bc_private.b.allocated : 0)); - base = ip->i_df.if_u1.if_extents; - ep = &base[idx]; + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_set_startblock(ep, NULLSTARTBLOCK((int)temp)); xfs_bmap_trace_post_update(fname, "RF", ip, idx, XFS_DATA_FORK); *dnew = temp; @@ -1149,7 +1129,7 @@ xfs_bmap_add_extent_delay_real( r[1].br_blockcount = temp2; xfs_bmap_trace_insert(fname, "0", ip, idx + 1, 2, &r[0], &r[1], XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx + 1, 2, &r[0], XFS_DATA_FORK); + xfs_iext_insert(ifp, idx + 1, 2, &r[0]); ip->i_df.if_lastex = idx + 1; ip->i_d.di_nextents++; if (cur == NULL) @@ -1204,13 +1184,13 @@ xfs_bmap_add_extent_delay_real( } } } - base = ip->i_df.if_u1.if_extents; - ep = &base[idx]; + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_set_startblock(ep, NULLSTARTBLOCK((int)temp)); xfs_bmap_trace_post_update(fname, "0", ip, idx, XFS_DATA_FORK); xfs_bmap_trace_pre_update(fname, "0", ip, idx + 2, XFS_DATA_FORK); - xfs_bmbt_set_startblock(ep + 2, NULLSTARTBLOCK((int)temp2)); + xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, idx + 2), + NULLSTARTBLOCK((int)temp2)); xfs_bmap_trace_post_update(fname, "0", ip, idx + 2, XFS_DATA_FORK); *dnew = temp + temp2; @@ -1254,10 +1234,9 @@ xfs_bmap_add_extent_unwritten_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t **curp, /* if *curp is null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp) /* inode logging flags */ { - xfs_bmbt_rec_t *base; /* base of extent entry list */ xfs_btree_cur_t *cur; /* btree cursor */ xfs_bmbt_rec_t *ep; /* extent entry for idx */ int error; /* error return value */ @@ -1265,6 +1244,7 @@ xfs_bmap_add_extent_unwritten_real( static char fname[] = "xfs_bmap_add_extent_unwritten_real"; #endif int i; /* temp state */ + xfs_ifork_t *ifp; /* inode fork pointer */ xfs_fileoff_t new_endoff; /* end offset of new entry */ xfs_exntst_t newext; /* new extent state */ xfs_exntst_t oldext; /* old extent state */ @@ -1298,8 +1278,8 @@ xfs_bmap_add_extent_unwritten_real( */ error = 0; cur = *curp; - base = ip->i_df.if_u1.if_extents; - ep = &base[idx]; + ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK); + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_get_all(ep, &PREV); newext = new->br_state; oldext = (newext == XFS_EXT_UNWRITTEN) ? @@ -1320,7 +1300,7 @@ xfs_bmap_add_extent_unwritten_real( * Don't set contiguous if the combined extent would be too large. */ if (STATE_SET_TEST(LEFT_VALID, idx > 0)) { - xfs_bmbt_get_all(ep - 1, &LEFT); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx - 1), &LEFT); STATE_SET(LEFT_DELAY, ISNULLSTARTBLOCK(LEFT.br_startblock)); } STATE_SET(LEFT_CONTIG, @@ -1337,7 +1317,7 @@ xfs_bmap_add_extent_unwritten_real( if (STATE_SET_TEST(RIGHT_VALID, idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1)) { - xfs_bmbt_get_all(ep + 1, &RIGHT); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx + 1), &RIGHT); STATE_SET(RIGHT_DELAY, ISNULLSTARTBLOCK(RIGHT.br_startblock)); } STATE_SET(RIGHT_CONTIG, @@ -1363,14 +1343,14 @@ xfs_bmap_add_extent_unwritten_real( */ xfs_bmap_trace_pre_update(fname, "LF|RF|LC|RC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + PREV.br_blockcount + RIGHT.br_blockcount); xfs_bmap_trace_post_update(fname, "LF|RF|LC|RC", ip, idx - 1, XFS_DATA_FORK); xfs_bmap_trace_delete(fname, "LF|RF|LC|RC", ip, idx, 2, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx, 2, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx, 2); ip->i_df.if_lastex = idx - 1; ip->i_d.di_nextents -= 2; if (cur == NULL) @@ -1409,14 +1389,14 @@ xfs_bmap_add_extent_unwritten_real( */ xfs_bmap_trace_pre_update(fname, "LF|RF|LC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + PREV.br_blockcount); xfs_bmap_trace_post_update(fname, "LF|RF|LC", ip, idx - 1, XFS_DATA_FORK); ip->i_df.if_lastex = idx - 1; xfs_bmap_trace_delete(fname, "LF|RF|LC", ip, idx, 1, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx, 1, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx, 1); ip->i_d.di_nextents--; if (cur == NULL) rval = XFS_ILOG_CORE | XFS_ILOG_DEXT; @@ -1456,7 +1436,7 @@ xfs_bmap_add_extent_unwritten_real( ip->i_df.if_lastex = idx; xfs_bmap_trace_delete(fname, "LF|RF|RC", ip, idx + 1, 1, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx + 1, 1, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx + 1, 1); ip->i_d.di_nextents--; if (cur == NULL) rval = XFS_ILOG_CORE | XFS_ILOG_DEXT; @@ -1516,7 +1496,7 @@ xfs_bmap_add_extent_unwritten_real( */ xfs_bmap_trace_pre_update(fname, "LF|LC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), LEFT.br_blockcount + new->br_blockcount); xfs_bmbt_set_startoff(ep, PREV.br_startoff + new->br_blockcount); @@ -1571,7 +1551,7 @@ xfs_bmap_add_extent_unwritten_real( xfs_bmap_trace_post_update(fname, "LF", ip, idx, XFS_DATA_FORK); xfs_bmap_trace_insert(fname, "LF", ip, idx, 1, new, NULL, XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx, 1, new, XFS_DATA_FORK); + xfs_iext_insert(ifp, idx, 1, new); ip->i_df.if_lastex = idx; ip->i_d.di_nextents++; if (cur == NULL) @@ -1609,7 +1589,8 @@ xfs_bmap_add_extent_unwritten_real( PREV.br_blockcount - new->br_blockcount); xfs_bmap_trace_post_update(fname, "RF|RC", ip, idx, XFS_DATA_FORK); - xfs_bmbt_set_allf(ep + 1, new->br_startoff, new->br_startblock, + xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, idx + 1), + new->br_startoff, new->br_startblock, new->br_blockcount + RIGHT.br_blockcount, newext); xfs_bmap_trace_post_update(fname, "RF|RC", ip, idx + 1, XFS_DATA_FORK); @@ -1649,7 +1630,7 @@ xfs_bmap_add_extent_unwritten_real( xfs_bmap_trace_post_update(fname, "RF", ip, idx, XFS_DATA_FORK); xfs_bmap_trace_insert(fname, "RF", ip, idx + 1, 1, new, NULL, XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx + 1, 1, new, XFS_DATA_FORK); + xfs_iext_insert(ifp, idx + 1, 1, new); ip->i_df.if_lastex = idx + 1; ip->i_d.di_nextents++; if (cur == NULL) @@ -1696,7 +1677,7 @@ xfs_bmap_add_extent_unwritten_real( r[1].br_state = oldext; xfs_bmap_trace_insert(fname, "0", ip, idx + 1, 2, &r[0], &r[1], XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx + 1, 2, &r[0], XFS_DATA_FORK); + xfs_iext_insert(ifp, idx + 1, 2, &r[0]); ip->i_df.if_lastex = idx + 1; ip->i_d.di_nextents += 2; if (cur == NULL) @@ -1770,15 +1751,15 @@ xfs_bmap_add_extent_hole_delay( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp, /* inode logging flags */ int rsvd) /* OK to allocate reserved blocks */ { - xfs_bmbt_rec_t *base; /* base of extent entry list */ - xfs_bmbt_rec_t *ep; /* extent list entry for idx */ + xfs_bmbt_rec_t *ep; /* extent record for idx */ #ifdef XFS_BMAP_TRACE static char fname[] = "xfs_bmap_add_extent_hole_delay"; #endif + xfs_ifork_t *ifp; /* inode fork pointer */ xfs_bmbt_irec_t left; /* left neighbor extent entry */ xfs_filblks_t newlen=0; /* new indirect size */ xfs_filblks_t oldlen=0; /* old indirect size */ @@ -1799,15 +1780,15 @@ xfs_bmap_add_extent_hole_delay( ((state &= ~MASK(b)), 0)) #define SWITCH_STATE (state & MASK2(LEFT_CONTIG, RIGHT_CONTIG)) - base = ip->i_df.if_u1.if_extents; - ep = &base[idx]; + ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK); + ep = xfs_iext_get_ext(ifp, idx); state = 0; ASSERT(ISNULLSTARTBLOCK(new->br_startblock)); /* * Check and set flags if this segment has a left neighbor */ if (STATE_SET_TEST(LEFT_VALID, idx > 0)) { - xfs_bmbt_get_all(ep - 1, &left); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx - 1), &left); STATE_SET(LEFT_DELAY, ISNULLSTARTBLOCK(left.br_startblock)); } /* @@ -1844,23 +1825,24 @@ xfs_bmap_add_extent_hole_delay( /* * New allocation is contiguous with delayed allocations * on the left and on the right. - * Merge all three into a single extent list entry. + * Merge all three into a single extent record. */ temp = left.br_blockcount + new->br_blockcount + right.br_blockcount; xfs_bmap_trace_pre_update(fname, "LC|RC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, temp); + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), temp); oldlen = STARTBLOCKVAL(left.br_startblock) + STARTBLOCKVAL(new->br_startblock) + STARTBLOCKVAL(right.br_startblock); newlen = xfs_bmap_worst_indlen(ip, temp); - xfs_bmbt_set_startblock(ep - 1, NULLSTARTBLOCK((int)newlen)); + xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, idx - 1), + NULLSTARTBLOCK((int)newlen)); xfs_bmap_trace_post_update(fname, "LC|RC", ip, idx - 1, XFS_DATA_FORK); xfs_bmap_trace_delete(fname, "LC|RC", ip, idx, 1, XFS_DATA_FORK); - xfs_bmap_delete_exlist(ip, idx, 1, XFS_DATA_FORK); + xfs_iext_remove(ifp, idx, 1); ip->i_df.if_lastex = idx - 1; break; @@ -1873,11 +1855,12 @@ xfs_bmap_add_extent_hole_delay( temp = left.br_blockcount + new->br_blockcount; xfs_bmap_trace_pre_update(fname, "LC", ip, idx - 1, XFS_DATA_FORK); - xfs_bmbt_set_blockcount(ep - 1, temp); + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), temp); oldlen = STARTBLOCKVAL(left.br_startblock) + STARTBLOCKVAL(new->br_startblock); newlen = xfs_bmap_worst_indlen(ip, temp); - xfs_bmbt_set_startblock(ep - 1, NULLSTARTBLOCK((int)newlen)); + xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, idx - 1), + NULLSTARTBLOCK((int)newlen)); xfs_bmap_trace_post_update(fname, "LC", ip, idx - 1, XFS_DATA_FORK); ip->i_df.if_lastex = idx - 1; @@ -1909,7 +1892,7 @@ xfs_bmap_add_extent_hole_delay( oldlen = newlen = 0; xfs_bmap_trace_insert(fname, "0", ip, idx, 1, new, NULL, XFS_DATA_FORK); - xfs_bmap_insert_exlist(ip, idx, 1, new, XFS_DATA_FORK); + xfs_iext_insert(ifp, idx, 1, new); ip->i_df.if_lastex = idx; break; } @@ -1940,7 +1923,7 @@ xfs_bmap_add_extent_hole_real( xfs_inode_t *ip, /* incore inode pointer */ xfs_extnum_t idx, /* extent number to update/insert */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *new, /* new data to put in extent list */ + xfs_bmbt_irec_t *new, /* new data to add to file extents */ int *logflagsp, /* inode logging flags */ int whichfork) /* data or attr fork */ { @@ -1970,13 +1953,13 @@ xfs_bmap_add_extent_hole_real( ifp = XFS_IFORK_PTR(ip, whichfork); ASSERT(idx <= ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)); - ep = &ifp->if_u1.if_extents[idx]; + ep = xfs_iext_get_ext(ifp, idx); state = 0; /* * Check and set flags if this segment has a left neighbor. */ if (STATE_SET_TEST(LEFT_VALID, idx > 0)) { - xfs_bmbt_get_all(ep - 1, &left); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx - 1), &left); STATE_SET(LEFT_DELAY, ISNULLSTARTBLOCK(left.br_startblock)); } /* @@ -2019,18 +2002,18 @@ xfs_bmap_add_extent_hole_real( /* * New allocation is contiguous with real allocations on the * left and on the right. - * Merge all three into a single extent list entry. + * Merge all three into a single extent record. */ xfs_bmap_trace_pre_update(fname, "LC|RC", ip, idx - 1, whichfork); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), left.br_blockcount + new->br_blockcount + right.br_blockcount); xfs_bmap_trace_post_update(fname, "LC|RC", ip, idx - 1, whichfork); xfs_bmap_trace_delete(fname, "LC|RC", ip, idx, 1, whichfork); - xfs_bmap_delete_exlist(ip, idx, 1, whichfork); + xfs_iext_remove(ifp, idx, 1); ifp->if_lastex = idx - 1; XFS_IFORK_NEXT_SET(ip, whichfork, XFS_IFORK_NEXTENTS(ip, whichfork) - 1); @@ -2062,7 +2045,7 @@ xfs_bmap_add_extent_hole_real( * Merge the new allocation with the left neighbor. */ xfs_bmap_trace_pre_update(fname, "LC", ip, idx - 1, whichfork); - xfs_bmbt_set_blockcount(ep - 1, + xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, idx - 1), left.br_blockcount + new->br_blockcount); xfs_bmap_trace_post_update(fname, "LC", ip, idx - 1, whichfork); ifp->if_lastex = idx - 1; @@ -2116,7 +2099,7 @@ xfs_bmap_add_extent_hole_real( */ xfs_bmap_trace_insert(fname, "0", ip, idx, 1, new, NULL, whichfork); - xfs_bmap_insert_exlist(ip, idx, 1, new, whichfork); + xfs_iext_insert(ifp, idx, 1, new); ifp->if_lastex = idx; XFS_IFORK_NEXT_SET(ip, whichfork, XFS_IFORK_NEXTENTS(ip, whichfork) + 1); @@ -2811,7 +2794,7 @@ xfs_bmap_alloc( /* * Transform a btree format file with only one leaf node, where the * extents list will fit in the inode, into an extents format file. - * Since the extent list is already in-core, all we have to do is + * Since the file extents are already in-core, all we have to do is * give up the space for the btree root and pitch the leaf block. */ STATIC int /* error */ @@ -2868,7 +2851,7 @@ xfs_bmap_btree_to_extents( } /* - * Called by xfs_bmapi to update extent list structure and the btree + * Called by xfs_bmapi to update file extent records and the btree * after removing space (or undoing a delayed allocation). */ STATIC int /* error */ @@ -2878,7 +2861,7 @@ xfs_bmap_del_extent( xfs_extnum_t idx, /* extent number to update/delete */ xfs_bmap_free_t *flist, /* list of extents to be freed */ xfs_btree_cur_t *cur, /* if null, not a btree */ - xfs_bmbt_irec_t *del, /* data to remove from extent list */ + xfs_bmbt_irec_t *del, /* data to remove from extents */ int *logflagsp, /* inode logging flags */ int whichfork, /* data or attr fork */ int rsvd) /* OK to allocate reserved blocks */ @@ -2903,7 +2886,6 @@ xfs_bmap_del_extent( xfs_filblks_t nblks; /* quota/sb block count */ xfs_bmbt_irec_t new; /* new record to be inserted */ /* REFERENCED */ - xfs_extnum_t nextents; /* number of extents in list */ uint qfield; /* quota field to update */ xfs_filblks_t temp; /* for indirect length calculations */ xfs_filblks_t temp2; /* for indirect length calculations */ @@ -2911,10 +2893,10 @@ xfs_bmap_del_extent( XFS_STATS_INC(xs_del_exlist); mp = ip->i_mount; ifp = XFS_IFORK_PTR(ip, whichfork); - nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - ASSERT(idx >= 0 && idx < nextents); + ASSERT((idx >= 0) && (idx < ifp->if_bytes / + (uint)sizeof(xfs_bmbt_rec_t))); ASSERT(del->br_blockcount > 0); - ep = &ifp->if_u1.if_extents[idx]; + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_get_all(ep, &got); ASSERT(got.br_startoff <= del->br_startoff); del_endoff = del->br_startoff + del->br_blockcount; @@ -2990,7 +2972,7 @@ xfs_bmap_del_extent( * Matches the whole extent. Delete the entry. */ xfs_bmap_trace_delete(fname, "3", ip, idx, 1, whichfork); - xfs_bmap_delete_exlist(ip, idx, 1, whichfork); + xfs_iext_remove(ifp, idx, 1); ifp->if_lastex = idx; if (delay) break; @@ -3160,7 +3142,7 @@ xfs_bmap_del_extent( xfs_bmap_trace_post_update(fname, "0", ip, idx, whichfork); xfs_bmap_trace_insert(fname, "0", ip, idx + 1, 1, &new, NULL, whichfork); - xfs_bmap_insert_exlist(ip, idx + 1, 1, &new, whichfork); + xfs_iext_insert(ifp, idx + 1, 1, &new); ifp->if_lastex = idx + 1; break; } @@ -3212,31 +3194,6 @@ xfs_bmap_del_free( kmem_zone_free(xfs_bmap_free_item_zone, free); } -/* - * Remove count entries from the extents array for inode "ip", starting - * at index "idx". Copies the remaining items down over the deleted ones, - * and gives back the excess memory. - */ -STATIC void -xfs_bmap_delete_exlist( - xfs_inode_t *ip, /* incore inode pointer */ - xfs_extnum_t idx, /* starting delete index */ - xfs_extnum_t count, /* count of items to delete */ - int whichfork) /* data or attr fork */ -{ - xfs_bmbt_rec_t *base; /* base of extent list */ - xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_extnum_t nextents; /* number of extents in list after */ - - ifp = XFS_IFORK_PTR(ip, whichfork); - ASSERT(ifp->if_flags & XFS_IFEXTENTS); - base = ifp->if_u1.if_extents; - nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - count; - memmove(&base[idx], &base[idx + count], - (nextents - idx) * sizeof(*base)); - xfs_iext_realloc(ip, -count, whichfork); -} - /* * Convert an extents-format file into a btree-format file. * The new file will have a root block (in the inode) and a single child block. @@ -3258,13 +3215,13 @@ xfs_bmap_extents_to_btree( xfs_bmbt_rec_t *arp; /* child record pointer */ xfs_bmbt_block_t *block; /* btree root block */ xfs_btree_cur_t *cur; /* bmap btree cursor */ - xfs_bmbt_rec_t *ep; /* extent list pointer */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ int error; /* error return value */ - xfs_extnum_t i, cnt; /* extent list index */ + xfs_extnum_t i, cnt; /* extent record index */ xfs_ifork_t *ifp; /* inode fork pointer */ xfs_bmbt_key_t *kp; /* root block key pointer */ xfs_mount_t *mp; /* mount structure */ - xfs_extnum_t nextents; /* extent list size */ + xfs_extnum_t nextents; /* number of file extents */ xfs_bmbt_ptr_t *pp; /* root block address pointer */ ifp = XFS_IFORK_PTR(ip, whichfork); @@ -3343,7 +3300,8 @@ xfs_bmap_extents_to_btree( ablock->bb_rightsib = cpu_to_be64(NULLDFSBNO); arp = XFS_BMAP_REC_IADDR(ablock, 1, cur); nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - for (ep = ifp->if_u1.if_extents, cnt = i = 0; i < nextents; i++, ep++) { + for (cnt = i = 0; i < nextents; i++) { + ep = xfs_iext_get_ext(ifp, i); if (!ISNULLSTARTBLOCK(xfs_bmbt_get_startblock(ep))) { arp->l0 = INT_GET(ep->l0, ARCH_CONVERT); arp->l1 = INT_GET(ep->l1, ARCH_CONVERT); @@ -3372,34 +3330,6 @@ xfs_bmap_extents_to_btree( return 0; } -/* - * Insert new item(s) in the extent list for inode "ip". - * Count new items are inserted at offset idx. - */ -STATIC void -xfs_bmap_insert_exlist( - xfs_inode_t *ip, /* incore inode pointer */ - xfs_extnum_t idx, /* starting index of new items */ - xfs_extnum_t count, /* number of inserted items */ - xfs_bmbt_irec_t *new, /* items to insert */ - int whichfork) /* data or attr fork */ -{ - xfs_bmbt_rec_t *base; /* extent list base */ - xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_extnum_t nextents; /* extent list size */ - xfs_extnum_t to; /* extent list index */ - - ifp = XFS_IFORK_PTR(ip, whichfork); - ASSERT(ifp->if_flags & XFS_IFEXTENTS); - xfs_iext_realloc(ip, count, whichfork); - base = ifp->if_u1.if_extents; - nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - memmove(&base[idx + count], &base[idx], - (nextents - (idx + count)) * sizeof(*base)); - for (to = idx; to < idx + count; to++, new++) - xfs_bmbt_set_all(&base[to], new); -} - /* * Helper routine to reset inode di_forkoff field when switching * attribute fork from local to extent format - we reset it where @@ -3457,8 +3387,8 @@ xfs_bmap_local_to_extents( error = 0; if (ifp->if_bytes) { xfs_alloc_arg_t args; /* allocation arguments */ - xfs_buf_t *bp; /* buffer for extent list block */ - xfs_bmbt_rec_t *ep; /* extent list pointer */ + xfs_buf_t *bp; /* buffer for extent block */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ args.tp = tp; args.mp = ip->i_mount; @@ -3492,8 +3422,8 @@ xfs_bmap_local_to_extents( xfs_trans_log_buf(tp, bp, 0, ifp->if_bytes - 1); xfs_bmap_forkoff_reset(args.mp, ip, whichfork); xfs_idata_realloc(ip, -ifp->if_bytes, whichfork); - xfs_iext_realloc(ip, 1, whichfork); - ep = ifp->if_u1.if_extents; + xfs_iext_add(ifp, 0, 1); + ep = xfs_iext_get_ext(ifp, 0); xfs_bmbt_set_allf(ep, 0, args.fsbno, 1, XFS_EXT_NORM); xfs_bmap_trace_post_update(fname, "new", ip, 0, whichfork); XFS_IFORK_NEXT_SET(ip, whichfork, 1); @@ -3518,7 +3448,7 @@ xfs_bmbt_rec_t * /* pointer to found extent entry */ xfs_bmap_do_search_extents( xfs_bmbt_rec_t *base, /* base of extent list */ xfs_extnum_t lastx, /* last extent index used */ - xfs_extnum_t nextents, /* extent list size */ + xfs_extnum_t nextents, /* number of file extents */ xfs_fileoff_t bno, /* block number searched for */ int *eofp, /* out: end of file found */ xfs_extnum_t *lastxp, /* out: last extent index */ @@ -3569,9 +3499,9 @@ xfs_bmap_do_search_extents( got.br_blockcount = xfs_bmbt_get_blockcount(ep); *eofp = 0; } else { - /* binary search the extents array */ low = 0; high = nextents - 1; + /* binary search the extents array */ while (low <= high) { XFS_STATS_INC(xs_cmp_exlist); lastx = (low + high) >> 1; @@ -3641,8 +3571,8 @@ xfs_bmap_search_extents( xfs_ifork_t *ifp; /* inode fork pointer */ xfs_bmbt_rec_t *base; /* base of extent list */ xfs_extnum_t lastx; /* last extent index used */ - xfs_extnum_t nextents; /* extent list size */ - xfs_bmbt_rec_t *ep; /* extent list entry pointer */ + xfs_extnum_t nextents; /* number of file extents */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ int rt; /* realtime flag */ XFS_STATS_INC(xs_look_exlist); @@ -3732,7 +3662,7 @@ xfs_bmap_trace_addentry( } /* - * Add bmap trace entry prior to a call to xfs_bmap_delete_exlist. + * Add bmap trace entry prior to a call to xfs_iext_remove. */ STATIC void xfs_bmap_trace_delete( @@ -3747,13 +3677,13 @@ xfs_bmap_trace_delete( ifp = XFS_IFORK_PTR(ip, whichfork); xfs_bmap_trace_addentry(XFS_BMAP_KTRACE_DELETE, fname, desc, ip, idx, - cnt, &ifp->if_u1.if_extents[idx], - cnt == 2 ? &ifp->if_u1.if_extents[idx + 1] : NULL, + cnt, xfs_iext_get_ext(ifp, idx), + cnt == 2 ? xfs_iext_get_ext(ifp, idx + 1) : NULL, whichfork); } /* - * Add bmap trace entry prior to a call to xfs_bmap_insert_exlist, or + * Add bmap trace entry prior to a call to xfs_iext_insert, or * reading in the extents list from the disk (in the btree). */ STATIC void @@ -3783,7 +3713,7 @@ xfs_bmap_trace_insert( } /* - * Add bmap trace entry after updating an extent list entry in place. + * Add bmap trace entry after updating an extent record in place. */ STATIC void xfs_bmap_trace_post_update( @@ -3797,11 +3727,11 @@ xfs_bmap_trace_post_update( ifp = XFS_IFORK_PTR(ip, whichfork); xfs_bmap_trace_addentry(XFS_BMAP_KTRACE_POST_UP, fname, desc, ip, idx, - 1, &ifp->if_u1.if_extents[idx], NULL, whichfork); + 1, xfs_iext_get_ext(ifp, idx), NULL, whichfork); } /* - * Add bmap trace entry prior to updating an extent list entry in place. + * Add bmap trace entry prior to updating an extent record in place. */ STATIC void xfs_bmap_trace_pre_update( @@ -3815,7 +3745,7 @@ xfs_bmap_trace_pre_update( ifp = XFS_IFORK_PTR(ip, whichfork); xfs_bmap_trace_addentry(XFS_BMAP_KTRACE_PRE_UP, fname, desc, ip, idx, 1, - &ifp->if_u1.if_extents[idx], NULL, whichfork); + xfs_iext_get_ext(ifp, idx), NULL, whichfork); } #endif /* XFS_BMAP_TRACE */ @@ -3892,7 +3822,7 @@ xfs_bmap_add_attrfork( int rsvd) /* xact may use reserved blks */ { xfs_fsblock_t firstblock; /* 1st block/ag allocated */ - xfs_bmap_free_t flist; /* freed extent list */ + xfs_bmap_free_t flist; /* freed extent records */ xfs_mount_t *mp; /* mount structure */ xfs_trans_t *tp; /* transaction pointer */ unsigned long s; /* spinlock spl value */ @@ -4146,7 +4076,7 @@ xfs_bmap_finish( xfs_efd_log_item_t *efd; /* extent free data */ xfs_efi_log_item_t *efi; /* extent free intention */ int error; /* error return value */ - xfs_bmap_free_item_t *free; /* free extent list item */ + xfs_bmap_free_item_t *free; /* free extent item */ unsigned int logres; /* new log reservation */ unsigned int logcount; /* new log count */ xfs_mount_t *mp; /* filesystem mount structure */ @@ -4242,9 +4172,9 @@ xfs_bmap_first_unused( xfs_fileoff_t *first_unused, /* unused block */ int whichfork) /* data or attr fork */ { - xfs_bmbt_rec_t *base; /* base of extent array */ xfs_bmbt_rec_t *ep; /* pointer to an extent entry */ int error; /* error return value */ + int idx; /* extent record index */ xfs_ifork_t *ifp; /* inode fork pointer */ xfs_fileoff_t lastaddr; /* last block number seen */ xfs_fileoff_t lowest; /* lowest useful block */ @@ -4265,10 +4195,8 @@ xfs_bmap_first_unused( return error; lowest = *first_unused; nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - base = &ifp->if_u1.if_extents[0]; - for (lastaddr = 0, max = lowest, ep = base; - ep < &base[nextents]; - ep++) { + for (idx = 0, lastaddr = 0, max = lowest; idx < nextents; idx++) { + ep = xfs_iext_get_ext(ifp, idx); off = xfs_bmbt_get_startoff(ep); /* * See if the hole before this extent will work. @@ -4287,8 +4215,8 @@ xfs_bmap_first_unused( /* * Returns the file-relative block number of the last block + 1 before * last_block (input value) in the file. - * This is not based on i_size, it is based on the extent list. - * Returns 0 for local files, as they do not have an extent list. + * This is not based on i_size, it is based on the extent records. + * Returns 0 for local files, as they do not have extent records. */ int /* error */ xfs_bmap_last_before( @@ -4335,8 +4263,8 @@ xfs_bmap_last_before( /* * Returns the file-relative block number of the first block past eof in - * the file. This is not based on i_size, it is based on the extent list. - * Returns 0 for local files, as they do not have an extent list. + * the file. This is not based on i_size, it is based on the extent records. + * Returns 0 for local files, as they do not have extent records. */ int /* error */ xfs_bmap_last_offset( @@ -4345,7 +4273,6 @@ xfs_bmap_last_offset( xfs_fileoff_t *last_block, /* last block */ int whichfork) /* data or attr fork */ { - xfs_bmbt_rec_t *base; /* base of extent array */ xfs_bmbt_rec_t *ep; /* pointer to last extent */ int error; /* error return value */ xfs_ifork_t *ifp; /* inode fork pointer */ @@ -4368,9 +4295,7 @@ xfs_bmap_last_offset( *last_block = 0; return 0; } - base = &ifp->if_u1.if_extents[0]; - ASSERT(base != NULL); - ep = &base[nextents - 1]; + ep = xfs_iext_get_ext(ifp, nextents - 1); *last_block = xfs_bmbt_get_startoff(ep) + xfs_bmbt_get_blockcount(ep); return 0; } @@ -4400,7 +4325,7 @@ xfs_bmap_one_block( return 0; ifp = XFS_IFORK_PTR(ip, whichfork); ASSERT(ifp->if_flags & XFS_IFEXTENTS); - ep = ifp->if_u1.if_extents; + ep = xfs_iext_get_ext(ifp, 0); xfs_bmbt_get_all(ep, &s); rval = s.br_startoff == 0 && s.br_blockcount == 1; if (rval && whichfork == XFS_DATA_FORK) @@ -4435,7 +4360,6 @@ xfs_bmap_read_extents( xfs_bmbt_ptr_t *pp; /* pointer to block address */ /* REFERENCED */ xfs_extnum_t room; /* number of entries there's room for */ - xfs_bmbt_rec_t *trp; /* target record pointer */ bno = NULLFSBLOCK; mp = ip->i_mount; @@ -4478,16 +4402,16 @@ xfs_bmap_read_extents( /* * Here with bp and block set to the leftmost leaf node in the tree. */ - room = ifp->if_bytes / (uint)sizeof(*trp); - trp = ifp->if_u1.if_extents; + room = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); i = 0; /* - * Loop over all leaf nodes. Copy information to the extent list. + * Loop over all leaf nodes. Copy information to the extent records. */ for (;;) { - xfs_bmbt_rec_t *frp, *temp; + xfs_bmbt_rec_t *frp, *trp; xfs_fsblock_t nextbno; xfs_extnum_t num_recs; + xfs_extnum_t start; num_recs = be16_to_cpu(block->bb_numrecs); @@ -4511,12 +4435,13 @@ xfs_bmap_read_extents( if (nextbno != NULLFSBLOCK) xfs_btree_reada_bufl(mp, nextbno, 1); /* - * Copy records into the extent list. + * Copy records into the extent records. */ frp = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, block, 1, mp->m_bmap_dmxr[0]); - temp = trp; - for (j = 0; j < num_recs; j++, frp++, trp++) { + start = i; + for (j = 0; j < num_recs; j++, i++, frp++) { + trp = xfs_iext_get_ext(ifp, i); trp->l0 = INT_GET(frp->l0, ARCH_CONVERT); trp->l1 = INT_GET(frp->l1, ARCH_CONVERT); } @@ -4526,14 +4451,14 @@ xfs_bmap_read_extents( * any "older" data bmap btree records for a * set bit in the "extent flag" position. */ - if (unlikely(xfs_check_nostate_extents(temp, num_recs))) { + if (unlikely(xfs_check_nostate_extents(ifp, + start, num_recs))) { XFS_ERROR_REPORT("xfs_bmap_read_extents(2)", XFS_ERRLEVEL_LOW, ip->i_mount); goto error0; } } - i += num_recs; xfs_trans_brelse(tp, bp); bno = nextbno; /* @@ -4546,7 +4471,7 @@ xfs_bmap_read_extents( return error; block = XFS_BUF_TO_BMBT_BLOCK(bp); } - ASSERT(i == ifp->if_bytes / (uint)sizeof(*trp)); + ASSERT(i == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))); ASSERT(i == XFS_IFORK_NEXTENTS(ip, whichfork)); xfs_bmap_trace_exlist(fname, ip, i, whichfork); return 0; @@ -4557,7 +4482,7 @@ error0: #ifdef XFS_BMAP_TRACE /* - * Add bmap trace insert entries for all the contents of the extent list. + * Add bmap trace insert entries for all the contents of the extent records. */ void xfs_bmap_trace_exlist( @@ -4566,16 +4491,15 @@ xfs_bmap_trace_exlist( xfs_extnum_t cnt, /* count of entries in the list */ int whichfork) /* data or attr fork */ { - xfs_bmbt_rec_t *base; /* base of extent list */ - xfs_bmbt_rec_t *ep; /* current entry in extent list */ - xfs_extnum_t idx; /* extent list entry number */ + xfs_bmbt_rec_t *ep; /* current extent record */ + xfs_extnum_t idx; /* extent record index */ xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_bmbt_irec_t s; /* extent list record */ + xfs_bmbt_irec_t s; /* file extent record */ ifp = XFS_IFORK_PTR(ip, whichfork); - ASSERT(cnt == ifp->if_bytes / (uint)sizeof(*base)); - base = ifp->if_u1.if_extents; - for (idx = 0, ep = base; idx < cnt; idx++, ep++) { + ASSERT(cnt == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))); + for (idx = 0; idx < cnt; idx++) { + ep = xfs_iext_get_ext(ifp, idx); xfs_bmbt_get_all(ep, &s); xfs_bmap_trace_insert(fname, "exlist", ip, idx, 1, &s, NULL, whichfork); @@ -4661,14 +4585,14 @@ xfs_bmapi( xfs_bmalloca_t bma; /* args for xfs_bmap_alloc */ xfs_btree_cur_t *cur; /* bmap btree cursor */ xfs_fileoff_t end; /* end of mapped file region */ - int eof; /* we've hit the end of extent list */ + int eof; /* we've hit the end of extents */ char contig; /* allocation must be one extent */ char delay; /* this request is for delayed alloc */ char exact; /* don't do all of wasdelayed extent */ char convert; /* unwritten extent I/O completion */ - xfs_bmbt_rec_t *ep; /* extent list entry pointer */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ int error; /* error return */ - xfs_bmbt_irec_t got; /* current extent list record */ + xfs_bmbt_irec_t got; /* current file extent record */ xfs_ifork_t *ifp; /* inode fork pointer */ xfs_extlen_t indlen; /* indirect blocks length */ xfs_extnum_t lastx; /* last useful extent number */ @@ -4680,7 +4604,7 @@ xfs_bmapi( int nallocs; /* number of extents alloc\'d */ xfs_extnum_t nextents; /* number of extents in file */ xfs_fileoff_t obno; /* old block number (offset) */ - xfs_bmbt_irec_t prev; /* previous extent list record */ + xfs_bmbt_irec_t prev; /* previous file extent record */ int tmp_logflags; /* temp flags holder */ int whichfork; /* data or attr fork */ char inhole; /* current location is hole in file */ @@ -4805,7 +4729,7 @@ xfs_bmapi( alen = (xfs_extlen_t)got.br_blockcount; aoff = got.br_startoff; if (lastx != NULLEXTNUM && lastx) { - ep = &ifp->if_u1.if_extents[lastx - 1]; + ep = xfs_iext_get_ext(ifp, lastx - 1); xfs_bmbt_get_all(ep, &prev); } } else if (wasdelay) { @@ -5016,7 +4940,7 @@ xfs_bmapi( if (error) goto error0; lastx = ifp->if_lastex; - ep = &ifp->if_u1.if_extents[lastx]; + ep = xfs_iext_get_ext(ifp, lastx); nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); xfs_bmbt_get_all(ep, &got); ASSERT(got.br_startoff <= aoff); @@ -5112,7 +5036,7 @@ xfs_bmapi( if (error) goto error0; lastx = ifp->if_lastex; - ep = &ifp->if_u1.if_extents[lastx]; + ep = xfs_iext_get_ext(ifp, lastx); nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); xfs_bmbt_get_all(ep, &got); /* @@ -5168,8 +5092,7 @@ xfs_bmapi( /* * Else go on to the next record. */ - ep++; - lastx++; + ep = xfs_iext_get_ext(ifp, ++lastx); if (lastx >= nextents) { eof = 1; prev = got; @@ -5199,7 +5122,7 @@ xfs_bmapi( error0: /* * Log everything. Do this after conversion, there's no point in - * logging the extent list if we've converted to btree format. + * logging the extent records if we've converted to btree format. */ if ((logflags & XFS_ILOG_FEXT(whichfork)) && XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS) @@ -5252,12 +5175,12 @@ xfs_bmapi_single( xfs_fsblock_t *fsb, /* output: mapped block */ xfs_fileoff_t bno) /* starting file offs. mapped */ { - int eof; /* we've hit the end of extent list */ + int eof; /* we've hit the end of extents */ int error; /* error return */ - xfs_bmbt_irec_t got; /* current extent list record */ + xfs_bmbt_irec_t got; /* current file extent record */ xfs_ifork_t *ifp; /* inode fork pointer */ xfs_extnum_t lastx; /* last useful extent number */ - xfs_bmbt_irec_t prev; /* previous extent list record */ + xfs_bmbt_irec_t prev; /* previous file extent record */ ifp = XFS_IFORK_PTR(ip, whichfork); if (unlikely( @@ -5312,18 +5235,18 @@ xfs_bunmapi( xfs_btree_cur_t *cur; /* bmap btree cursor */ xfs_bmbt_irec_t del; /* extent being deleted */ int eof; /* is deleting at eof */ - xfs_bmbt_rec_t *ep; /* extent list entry pointer */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ int error; /* error return value */ xfs_extnum_t extno; /* extent number in list */ - xfs_bmbt_irec_t got; /* current extent list entry */ + xfs_bmbt_irec_t got; /* current extent record */ xfs_ifork_t *ifp; /* inode fork pointer */ int isrt; /* freeing in rt area */ xfs_extnum_t lastx; /* last extent index used */ int logflags; /* transaction logging flags */ xfs_extlen_t mod; /* rt extent offset */ xfs_mount_t *mp; /* mount structure */ - xfs_extnum_t nextents; /* size of extent list */ - xfs_bmbt_irec_t prev; /* previous extent list entry */ + xfs_extnum_t nextents; /* number of file extents */ + xfs_bmbt_irec_t prev; /* previous extent record */ xfs_fileoff_t start; /* first file offset deleted */ int tmp_logflags; /* partial logging flags */ int wasdel; /* was a delayed alloc extent */ @@ -5369,7 +5292,7 @@ xfs_bunmapi( * file, back up to the last block if so... */ if (eof) { - ep = &ifp->if_u1.if_extents[--lastx]; + ep = xfs_iext_get_ext(ifp, --lastx); xfs_bmbt_get_all(ep, &got); bno = got.br_startoff + got.br_blockcount - 1; } @@ -5393,7 +5316,7 @@ xfs_bunmapi( if (got.br_startoff > bno) { if (--lastx < 0) break; - ep--; + ep = xfs_iext_get_ext(ifp, lastx); xfs_bmbt_get_all(ep, &got); } /* @@ -5440,7 +5363,8 @@ xfs_bunmapi( del.br_blockcount : mod; if (bno < got.br_startoff) { if (--lastx >= 0) - xfs_bmbt_get_all(--ep, &got); + xfs_bmbt_get_all(xfs_iext_get_ext( + ifp, lastx), &got); } continue; } @@ -5500,7 +5424,8 @@ xfs_bunmapi( * try again. */ ASSERT(lastx > 0); - xfs_bmbt_get_all(ep - 1, &prev); + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, + lastx - 1), &prev); ASSERT(prev.br_state == XFS_EXT_NORM); ASSERT(!ISNULLSTARTBLOCK(prev.br_startblock)); ASSERT(del.br_startblock == @@ -5587,12 +5512,12 @@ nodelete: * If not done go on to the next (previous) record. * Reset ep in case the extents array was re-alloced. */ - ep = &ifp->if_u1.if_extents[lastx]; + ep = xfs_iext_get_ext(ifp, lastx); if (bno != (xfs_fileoff_t)-1 && bno >= start) { if (lastx >= XFS_IFORK_NEXTENTS(ip, whichfork) || xfs_bmbt_get_startoff(ep) > bno) { - lastx--; - ep--; + if (--lastx >= 0) + ep = xfs_iext_get_ext(ifp, lastx); } if (lastx >= 0) xfs_bmbt_get_all(ep, &got); @@ -5636,7 +5561,7 @@ nodelete: error0: /* * Log everything. Do this after conversion, there's no point in - * logging the extent list if we've converted to btree format. + * logging the extent records if we've converted to btree format. */ if ((logflags & XFS_ILOG_FEXT(whichfork)) && XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS) @@ -5892,9 +5817,9 @@ xfs_bmap_isaeof( { int error; /* error return value */ xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_bmbt_rec_t *lastrec; /* extent list entry pointer */ - xfs_extnum_t nextents; /* size of extent list */ - xfs_bmbt_irec_t s; /* expanded extent list entry */ + xfs_bmbt_rec_t *lastrec; /* extent record pointer */ + xfs_extnum_t nextents; /* number of file extents */ + xfs_bmbt_irec_t s; /* expanded extent record */ ASSERT(whichfork == XFS_DATA_FORK); ifp = XFS_IFORK_PTR(ip, whichfork); @@ -5909,7 +5834,7 @@ xfs_bmap_isaeof( /* * Go to the last extent */ - lastrec = &ifp->if_u1.if_extents[nextents - 1]; + lastrec = xfs_iext_get_ext(ifp, nextents - 1); xfs_bmbt_get_all(lastrec, &s); /* * Check we are allocating in the last extent (for delayed allocations) @@ -5936,8 +5861,8 @@ xfs_bmap_eof( xfs_fsblock_t blockcount; /* extent block count */ int error; /* error return value */ xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_bmbt_rec_t *lastrec; /* extent list entry pointer */ - xfs_extnum_t nextents; /* size of extent list */ + xfs_bmbt_rec_t *lastrec; /* extent record pointer */ + xfs_extnum_t nextents; /* number of file extents */ xfs_fileoff_t startoff; /* extent starting file offset */ ASSERT(whichfork == XFS_DATA_FORK); @@ -5953,7 +5878,7 @@ xfs_bmap_eof( /* * Go to the last extent */ - lastrec = &ifp->if_u1.if_extents[nextents - 1]; + lastrec = xfs_iext_get_ext(ifp, nextents - 1); startoff = xfs_bmbt_get_startoff(lastrec); blockcount = xfs_bmbt_get_blockcount(lastrec); *eof = endoff >= startoff + blockcount; @@ -5969,18 +5894,21 @@ xfs_bmap_check_extents( xfs_inode_t *ip, /* incore inode pointer */ int whichfork) /* data or attr fork */ { - xfs_bmbt_rec_t *base; /* base of extents list */ xfs_bmbt_rec_t *ep; /* current extent entry */ + xfs_extnum_t idx; /* extent record index */ xfs_ifork_t *ifp; /* inode fork pointer */ xfs_extnum_t nextents; /* number of extents in list */ + xfs_bmbt_rec_t *nextp; /* next extent entry */ ifp = XFS_IFORK_PTR(ip, whichfork); ASSERT(ifp->if_flags & XFS_IFEXTENTS); - base = ifp->if_u1.if_extents; nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - for (ep = base; ep < &base[nextents - 1]; ep++) { + ep = xfs_iext_get_ext(ifp, 0); + for (idx = 0; idx < nextents - 1; idx++) { + nextp = xfs_iext_get_ext(ifp, idx + 1); xfs_btree_check_rec(XFS_BTNUM_BMAP, (void *)ep, - (void *)(ep + 1)); + (void *)(nextp)); + ep = nextp; } } @@ -6119,12 +6047,14 @@ xfs_bmap_check_leaf_extents( xfs_fsblock_t bno; /* block # of "block" */ xfs_buf_t *bp; /* buffer for "block" */ int error; /* error return value */ - xfs_extnum_t i=0; /* index into the extents list */ + xfs_extnum_t i=0, j; /* index into the extents list */ xfs_ifork_t *ifp; /* fork structure */ int level; /* btree level, for checking */ xfs_mount_t *mp; /* file system mount structure */ xfs_bmbt_ptr_t *pp; /* pointer to block address */ - xfs_bmbt_rec_t *ep, *lastp; /* extent pointers in block entry */ + xfs_bmbt_rec_t *ep; /* pointer to current extent */ + xfs_bmbt_rec_t *lastp; /* pointer to previous extent */ + xfs_bmbt_rec_t *nextp; /* pointer to next extent */ int bp_release = 0; if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE) { @@ -6194,7 +6124,6 @@ xfs_bmap_check_leaf_extents( */ lastp = NULL; for (;;) { - xfs_bmbt_rec_t *frp; xfs_fsblock_t nextbno; xfs_extnum_t num_recs; @@ -6213,18 +6142,20 @@ xfs_bmap_check_leaf_extents( * conform with the first entry in this one. */ - frp = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, + ep = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, block, 1, mp->m_bmap_dmxr[0]); - - for (ep = frp;ep < frp + (num_recs - 1); ep++) { + for (j = 1; j < num_recs; j++) { + nextp = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, xfs_bmbt, + block, j + 1, mp->m_bmap_dmxr[0]); if (lastp) { xfs_btree_check_rec(XFS_BTNUM_BMAP, (void *)lastp, (void *)ep); } xfs_btree_check_rec(XFS_BTNUM_BMAP, (void *)ep, - (void *)(ep + 1)); + (void *)(nextp)); + lastp = ep; + ep = nextp; } - lastp = frp + num_recs - 1; /* For the next iteration */ i += num_recs; if (bp_release) { @@ -6288,7 +6219,7 @@ xfs_bmap_count_blocks( mp = ip->i_mount; ifp = XFS_IFORK_PTR(ip, whichfork); if ( XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS ) { - if (unlikely(xfs_bmap_count_leaves(ifp->if_u1.if_extents, + if (unlikely(xfs_bmap_count_leaves(ifp, 0, ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t), count) < 0)) { XFS_ERROR_REPORT("xfs_bmap_count_blocks(1)", @@ -6310,7 +6241,7 @@ xfs_bmap_count_blocks( ASSERT(XFS_FSB_TO_AGBNO(mp, INT_GET(*pp, ARCH_CONVERT)) < mp->m_sb.sb_agblocks); bno = INT_GET(*pp, ARCH_CONVERT); - if (unlikely(xfs_bmap_count_tree(mp, tp, bno, level, count) < 0)) { + if (unlikely(xfs_bmap_count_tree(mp, tp, ifp, bno, level, count) < 0)) { XFS_ERROR_REPORT("xfs_bmap_count_blocks(2)", XFS_ERRLEVEL_LOW, mp); return XFS_ERROR(EFSCORRUPTED); @@ -6327,6 +6258,7 @@ int /* error */ xfs_bmap_count_tree( xfs_mount_t *mp, /* file system mount point */ xfs_trans_t *tp, /* transaction pointer */ + xfs_ifork_t *ifp, /* inode fork pointer */ xfs_fsblock_t blockno, /* file system block number */ int levelin, /* level in btree */ int *count) /* Count of blocks */ @@ -6339,7 +6271,6 @@ xfs_bmap_count_tree( xfs_fsblock_t nextbno; xfs_bmbt_block_t *block, *nextblock; int numrecs; - xfs_bmbt_rec_t *frp; if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp, XFS_BMAP_BTREE_REF))) return error; @@ -6364,7 +6295,7 @@ xfs_bmap_count_tree( xfs_bmbt, block, 1, mp->m_bmap_dmxr[1]); bno = INT_GET(*pp, ARCH_CONVERT); if (unlikely((error = - xfs_bmap_count_tree(mp, tp, bno, level, count)) < 0)) { + xfs_bmap_count_tree(mp, tp, ifp, bno, level, count)) < 0)) { xfs_trans_brelse(tp, bp); XFS_ERROR_REPORT("xfs_bmap_count_tree(1)", XFS_ERRLEVEL_LOW, mp); @@ -6376,9 +6307,8 @@ xfs_bmap_count_tree( for (;;) { nextbno = be64_to_cpu(block->bb_rightsib); numrecs = be16_to_cpu(block->bb_numrecs); - frp = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, - xfs_bmbt, block, 1, mp->m_bmap_dmxr[0]); - if (unlikely(xfs_bmap_disk_count_leaves(frp, numrecs, count) < 0)) { + if (unlikely(xfs_bmap_disk_count_leaves(ifp, mp, + 0, block, numrecs, count) < 0)) { xfs_trans_brelse(tp, bp); XFS_ERROR_REPORT("xfs_bmap_count_tree(2)", XFS_ERRLEVEL_LOW, mp); @@ -6399,33 +6329,45 @@ xfs_bmap_count_tree( } /* - * Count leaf blocks given a pointer to an extent list. + * Count leaf blocks given a range of extent records. */ int xfs_bmap_count_leaves( - xfs_bmbt_rec_t *frp, + xfs_ifork_t *ifp, + xfs_extnum_t idx, int numrecs, int *count) { int b; + xfs_bmbt_rec_t *frp; - for ( b = 1; b <= numrecs; b++, frp++) + for (b = 0; b < numrecs; b++) { + frp = xfs_iext_get_ext(ifp, idx + b); *count += xfs_bmbt_get_blockcount(frp); + } return 0; } /* - * Count leaf blocks given a pointer to an extent list originally in btree format. + * Count leaf blocks given a range of extent records originally + * in btree format. */ int xfs_bmap_disk_count_leaves( - xfs_bmbt_rec_t *frp, + xfs_ifork_t *ifp, + xfs_mount_t *mp, + xfs_extnum_t idx, + xfs_bmbt_block_t *block, int numrecs, int *count) { int b; + xfs_bmbt_rec_t *frp; - for ( b = 1; b <= numrecs; b++, frp++) + for (b = 1; b <= numrecs; b++) { + frp = XFS_BTREE_REC_ADDR(mp->m_sb.sb_blocksize, + xfs_bmbt, block, idx + b, mp->m_bmap_dmxr[0]); *count += xfs_bmbt_disk_get_blockcount(frp); + } return 0; } -- cgit v1.2.3-70-g09d2 From 0293ce3a9fd1b34c933a96577a8ba737b681cf75 Mon Sep 17 00:00:00 2001 From: Mandy Kirkconnell Date: Tue, 14 Mar 2006 13:30:23 +1100 Subject: [XFS] 929045 567344 This mod introduces multi-level in-core file extent functionality, building upon the new layout introduced in mod xfs-linux:xfs-kern:207390a. The new multi-level extent allocations are only required for heavily fragmented files, so the old-style linear extent list is used on files until the extents reach a pre-determined size of 4k. 4k buffers are used because this is the system page size on Linux i386 and systems with larger page sizes don't seem to gain much, if anything, by using their native page size as the extent buffer size. Also, using 4k extent buffers everywhere provides a consistent interface for CXFS across different platforms. The 4k extent buffers are managed by an indirection array (xfs_ext_irec_t) which is basically just a pointer array with a bit of extra information to keep track of the number of extents in each buffer as well as the extent offset of each buffer. Major changes include: - Add multi-level in-core file extent functionality to the xfs_iext_ subroutines introduced in mod: xfs-linux:xfs-kern:207390a - Introduce 13 new subroutines which add functionality for multi-level in-core file extents: xfs_iext_add_indirect_multi() xfs_iext_remove_indirect() xfs_iext_realloc_indirect() xfs_iext_indirect_to_direct() xfs_iext_bno_to_irec() xfs_iext_idx_to_irec() xfs_iext_irec_init() xfs_iext_irec_new() xfs_iext_irec_remove() xfs_iext_irec_compact() xfs_iext_irec_compact_pages() xfs_iext_irec_compact_full() xfs_iext_irec_update_extoffs() SGI-PV: 928864 SGI-Modid: xfs-linux-melb:xfs-kern:207393a Signed-off-by: Mandy Kirkconnell Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 59 +++- fs/xfs/xfs_bmap.h | 18 ++ fs/xfs/xfs_bmap_btree.h | 8 - fs/xfs/xfs_inode.c | 715 +++++++++++++++++++++++++++++++++++++++++++++++- fs/xfs/xfs_inode.h | 52 +++- 5 files changed, 826 insertions(+), 26 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index 53c47a181f8..81a95b684b6 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -3423,6 +3423,7 @@ xfs_bmap_local_to_extents( xfs_bmap_forkoff_reset(args.mp, ip, whichfork); xfs_idata_realloc(ip, -ifp->if_bytes, whichfork); xfs_iext_add(ifp, 0, 1); + ASSERT((ifp->if_flags & (XFS_IFEXTENTS|XFS_IFEXTIREC)) == XFS_IFEXTENTS); ep = xfs_iext_get_ext(ifp, 0); xfs_bmbt_set_allf(ep, 0, args.fsbno, 1, XFS_EXT_NORM); xfs_bmap_trace_post_update(fname, "new", ip, 0, whichfork); @@ -3551,6 +3552,54 @@ xfs_bmap_do_search_extents( return ep; } +/* + * Call xfs_bmap_do_search_extents() to search for the extent + * record containing block bno. If in multi-level in-core extent + * allocation mode, find and extract the target extent buffer, + * otherwise just use the direct extent list. + */ +xfs_bmbt_rec_t * /* pointer to found extent entry */ +xfs_bmap_search_multi_extents( + xfs_ifork_t *ifp, /* inode fork pointer */ + xfs_fileoff_t bno, /* block number searched for */ + int *eofp, /* out: end of file found */ + xfs_extnum_t *lastxp, /* out: last extent index */ + xfs_bmbt_irec_t *gotp, /* out: extent entry found */ + xfs_bmbt_irec_t *prevp) /* out: previous extent entry found */ +{ + xfs_bmbt_rec_t *base; /* base of extent records */ + xfs_bmbt_rec_t *ep; /* extent record pointer */ + xfs_ext_irec_t *erp = NULL; /* indirection array pointer */ + xfs_extnum_t lastx; /* last extent index */ + xfs_extnum_t nextents; /* number of file extents */ + + /* + * For multi-level extent allocation mode, find the + * target extent list and pass only the contiguous + * list to xfs_bmap_do_search_extents. Convert lastx + * from a file extent index to an index within the + * target extent list. + */ + if (ifp->if_flags & XFS_IFEXTIREC) { + int erp_idx = 0; + erp = xfs_iext_bno_to_irec(ifp, bno, &erp_idx); + base = erp->er_extbuf; + nextents = erp->er_extcount; + lastx = ifp->if_lastex - erp->er_extoff; + } else { + base = &ifp->if_u1.if_extents[0]; + nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); + lastx = ifp->if_lastex; + } + ep = xfs_bmap_do_search_extents(base, lastx, nextents, bno, + eofp, lastxp, gotp, prevp); + /* Convert lastx back to file-based index */ + if (ifp->if_flags & XFS_IFEXTIREC) { + *lastxp += erp->er_extoff; + } + return ep; +} + /* * Search the extents list for the inode, for the extent containing bno. * If bno lies in a hole, point to the next entry. If bno lies past eof, @@ -3569,20 +3618,14 @@ xfs_bmap_search_extents( xfs_bmbt_irec_t *prevp) /* out: previous extent entry found */ { xfs_ifork_t *ifp; /* inode fork pointer */ - xfs_bmbt_rec_t *base; /* base of extent list */ - xfs_extnum_t lastx; /* last extent index used */ - xfs_extnum_t nextents; /* number of file extents */ xfs_bmbt_rec_t *ep; /* extent record pointer */ int rt; /* realtime flag */ XFS_STATS_INC(xs_look_exlist); ifp = XFS_IFORK_PTR(ip, whichfork); - lastx = ifp->if_lastex; - nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - base = &ifp->if_u1.if_extents[0]; - ep = xfs_bmap_do_search_extents(base, lastx, nextents, bno, eofp, - lastxp, gotp, prevp); + ep = xfs_bmap_search_multi_extents(ifp, bno, eofp, lastxp, gotp, prevp); + rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip); if (unlikely(!rt && !gotp->br_startblock && (*lastxp != NULLEXTNUM))) { cmn_err(CE_PANIC,"Access to block zero: fs: <%s> inode: %lld " diff --git a/fs/xfs/xfs_bmap.h b/fs/xfs/xfs_bmap.h index 4c05f95452c..011ccaa9a1c 100644 --- a/fs/xfs/xfs_bmap.h +++ b/fs/xfs/xfs_bmap.h @@ -352,6 +352,24 @@ xfs_check_nostate_extents( xfs_extnum_t idx, xfs_extnum_t num); +/* + * Call xfs_bmap_do_search_extents() to search for the extent + * record containing block bno. If in multi-level in-core extent + * allocation mode, find and extract the target extent buffer, + * otherwise just use the direct extent list. + */ +xfs_bmbt_rec_t * +xfs_bmap_search_multi_extents(struct xfs_ifork *, xfs_fileoff_t, int *, + xfs_extnum_t *, xfs_bmbt_irec_t *, xfs_bmbt_irec_t *); + +/* + * Search an extent list for the extent which includes block + * bno. + */ +xfs_bmbt_rec_t *xfs_bmap_do_search_extents(xfs_bmbt_rec_t *, + xfs_extnum_t, xfs_extnum_t, xfs_fileoff_t, int *, + xfs_extnum_t *, xfs_bmbt_irec_t *, xfs_bmbt_irec_t *); + #endif /* __KERNEL__ */ #endif /* __XFS_BMAP_H__ */ diff --git a/fs/xfs/xfs_bmap_btree.h b/fs/xfs/xfs_bmap_btree.h index e095a2d344a..6478cfa0e53 100644 --- a/fs/xfs/xfs_bmap_btree.h +++ b/fs/xfs/xfs_bmap_btree.h @@ -372,14 +372,6 @@ extern int xfs_bmbt_get_rec(struct xfs_btree_cur *, xfs_fileoff_t *, xfs_exntst_t *, int *); #endif -/* - * Search an extent list for the extent which includes block - * bno. - */ -xfs_bmbt_rec_t *xfs_bmap_do_search_extents(xfs_bmbt_rec_t *, - xfs_extnum_t, xfs_extnum_t, xfs_fileoff_t, int *, - xfs_extnum_t *, xfs_bmbt_irec_t *, xfs_bmbt_irec_t *); - #endif /* __KERNEL__ */ #endif /* __XFS_BMAP_BTREE_H__ */ diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index 6459395a0e4..580fa075803 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -2630,8 +2630,9 @@ xfs_idestroy_fork( ifp->if_real_bytes = 0; } } else if ((ifp->if_flags & XFS_IFEXTENTS) && - (ifp->if_u1.if_extents != NULL) && - (ifp->if_u1.if_extents != ifp->if_u2.if_inline_ext)) { + ((ifp->if_flags & XFS_IFEXTIREC) || + ((ifp->if_u1.if_extents != NULL) && + (ifp->if_u1.if_extents != ifp->if_u2.if_inline_ext)))) { ASSERT(ifp->if_real_bytes != 0); xfs_iext_destroy(ifp); } @@ -3622,7 +3623,16 @@ xfs_iext_get_ext( xfs_extnum_t idx) /* index of target extent */ { ASSERT(idx >= 0); - if (ifp->if_bytes) { + if ((ifp->if_flags & XFS_IFEXTIREC) && (idx == 0)) { + return ifp->if_u1.if_ext_irec->er_extbuf; + } else if (ifp->if_flags & XFS_IFEXTIREC) { + xfs_ext_irec_t *erp; /* irec pointer */ + int erp_idx = 0; /* irec index */ + xfs_extnum_t page_idx = idx; /* ext index in target list */ + + erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 0); + return &erp->er_extbuf[page_idx]; + } else if (ifp->if_bytes) { return &ifp->if_u1.if_extents[idx]; } else { return NULL; @@ -3691,6 +3701,7 @@ xfs_iext_add( } ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext; ifp->if_real_bytes = 0; + ifp->if_lastex = nextents + ext_diff; } /* * Otherwise use a linear (direct) extent list. @@ -3698,7 +3709,7 @@ xfs_iext_add( * xfs_iext_realloc_direct will switch us from * inline to direct extent allocation mode. */ - else { + else if (nextents + ext_diff <= XFS_LINEAR_EXTS) { xfs_iext_realloc_direct(ifp, new_size); if (idx < nextents) { memmove(&ifp->if_u1.if_extents[idx + ext_diff], @@ -3707,14 +3718,182 @@ xfs_iext_add( memset(&ifp->if_u1.if_extents[idx], 0, byte_diff); } } + /* Indirection array */ + else { + xfs_ext_irec_t *erp; + int erp_idx = 0; + int page_idx = idx; + + ASSERT(nextents + ext_diff > XFS_LINEAR_EXTS); + if (ifp->if_flags & XFS_IFEXTIREC) { + erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 1); + } else { + xfs_iext_irec_init(ifp); + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + erp = ifp->if_u1.if_ext_irec; + } + /* Extents fit in target extent page */ + if (erp && erp->er_extcount + ext_diff <= XFS_LINEAR_EXTS) { + if (page_idx < erp->er_extcount) { + memmove(&erp->er_extbuf[page_idx + ext_diff], + &erp->er_extbuf[page_idx], + (erp->er_extcount - page_idx) * + sizeof(xfs_bmbt_rec_t)); + memset(&erp->er_extbuf[page_idx], 0, byte_diff); + } + erp->er_extcount += ext_diff; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff); + } + /* Insert a new extent page */ + else if (erp) { + xfs_iext_add_indirect_multi(ifp, + erp_idx, page_idx, ext_diff); + } + /* + * If extent(s) are being appended to the last page in + * the indirection array and the new extent(s) don't fit + * in the page, then erp is NULL and erp_idx is set to + * the next index needed in the indirection array. + */ + else { + int count = ext_diff; + + while (count) { + erp = xfs_iext_irec_new(ifp, erp_idx); + erp->er_extcount = count; + count -= MIN(count, (int)XFS_LINEAR_EXTS); + if (count) { + erp_idx++; + } + } + } + } ifp->if_bytes = new_size; } +/* + * This is called when incore extents are being added to the indirection + * array and the new extents do not fit in the target extent list. The + * erp_idx parameter contains the irec index for the target extent list + * in the indirection array, and the idx parameter contains the extent + * index within the list. The number of extents being added is stored + * in the count parameter. + * + * |-------| |-------| + * | | | | idx - number of extents before idx + * | idx | | count | + * | | | | count - number of extents being inserted at idx + * |-------| |-------| + * | count | | nex2 | nex2 - number of extents after idx + count + * |-------| |-------| + */ +void +xfs_iext_add_indirect_multi( + xfs_ifork_t *ifp, /* inode fork pointer */ + int erp_idx, /* target extent irec index */ + xfs_extnum_t idx, /* index within target list */ + int count) /* new extents being added */ +{ + int byte_diff; /* new bytes being added */ + xfs_ext_irec_t *erp; /* pointer to irec entry */ + xfs_extnum_t ext_diff; /* number of extents to add */ + xfs_extnum_t ext_cnt; /* new extents still needed */ + xfs_extnum_t nex2; /* extents after idx + count */ + xfs_bmbt_rec_t *nex2_ep = NULL; /* temp list for nex2 extents */ + int nlists; /* number of irec's (lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + nex2 = erp->er_extcount - idx; + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + + /* + * Save second part of target extent list + * (all extents past */ + if (nex2) { + byte_diff = nex2 * sizeof(xfs_bmbt_rec_t); + nex2_ep = (xfs_bmbt_rec_t *) kmem_alloc(byte_diff, KM_SLEEP); + memmove(nex2_ep, &erp->er_extbuf[idx], byte_diff); + erp->er_extcount -= nex2; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -nex2); + memset(&erp->er_extbuf[idx], 0, byte_diff); + } + + /* + * Add the new extents to the end of the target + * list, then allocate new irec record(s) and + * extent buffer(s) as needed to store the rest + * of the new extents. + */ + ext_cnt = count; + ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS - erp->er_extcount); + if (ext_diff) { + erp->er_extcount += ext_diff; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff); + ext_cnt -= ext_diff; + } + while (ext_cnt) { + erp_idx++; + erp = xfs_iext_irec_new(ifp, erp_idx); + ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS); + erp->er_extcount = ext_diff; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff); + ext_cnt -= ext_diff; + } + + /* Add nex2 extents back to indirection array */ + if (nex2) { + xfs_extnum_t ext_avail; + int i; + + byte_diff = nex2 * sizeof(xfs_bmbt_rec_t); + ext_avail = XFS_LINEAR_EXTS - erp->er_extcount; + i = 0; + /* + * If nex2 extents fit in the current page, append + * nex2_ep after the new extents. + */ + if (nex2 <= ext_avail) { + i = erp->er_extcount; + } + /* + * Otherwise, check if space is available in the + * next page. + */ + else if ((erp_idx < nlists - 1) && + (nex2 <= (ext_avail = XFS_LINEAR_EXTS - + ifp->if_u1.if_ext_irec[erp_idx+1].er_extcount))) { + erp_idx++; + erp++; + /* Create a hole for nex2 extents */ + memmove(&erp->er_extbuf[nex2], erp->er_extbuf, + erp->er_extcount * sizeof(xfs_bmbt_rec_t)); + } + /* + * Final choice, create a new extent page for + * nex2 extents. + */ + else { + erp_idx++; + erp = xfs_iext_irec_new(ifp, erp_idx); + } + memmove(&erp->er_extbuf[i], nex2_ep, byte_diff); + kmem_free(nex2_ep, byte_diff); + erp->er_extcount += nex2; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, nex2); + } +} + /* * This is called when the amount of space required for incore file * extents needs to be decreased. The ext_diff parameter stores the * number of extents to be removed and the idx parameter contains * the extent index where the extents will be removed from. + * + * If the amount of space needed has decreased below the linear + * limit, XFS_IEXT_BUFSZ, then switch to using the contiguous + * extent array. Otherwise, use kmem_realloc() to adjust the + * size to what is needed. */ void xfs_iext_remove( @@ -3731,6 +3910,8 @@ xfs_iext_remove( if (new_size == 0) { xfs_iext_destroy(ifp); + } else if (ifp->if_flags & XFS_IFEXTIREC) { + xfs_iext_remove_indirect(ifp, idx, ext_diff); } else if (ifp->if_real_bytes) { xfs_iext_remove_direct(ifp, idx, ext_diff); } else { @@ -3751,6 +3932,7 @@ xfs_iext_remove_inline( { int nextents; /* number of extents in file */ + ASSERT(!(ifp->if_flags & XFS_IFEXTIREC)); ASSERT(idx < XFS_INLINE_EXTS); nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); ASSERT(((nextents - ext_diff) > 0) && @@ -3788,6 +3970,7 @@ xfs_iext_remove_direct( xfs_extnum_t nextents; /* number of extents in file */ int new_size; /* size of extents after removal */ + ASSERT(!(ifp->if_flags & XFS_IFEXTIREC)); new_size = ifp->if_bytes - (ext_diff * sizeof(xfs_bmbt_rec_t)); nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); @@ -3815,6 +3998,84 @@ xfs_iext_remove_direct( ifp->if_bytes = new_size; } +/* + * This is called when incore extents are being removed from the + * indirection array and the extents being removed span multiple extent + * buffers. The idx parameter contains the file extent index where we + * want to begin removing extents, and the count parameter contains + * how many extents need to be removed. + * + * |-------| |-------| + * | nex1 | | | nex1 - number of extents before idx + * |-------| | count | + * | | | | count - number of extents being removed at idx + * | count | |-------| + * | | | nex2 | nex2 - number of extents after idx + count + * |-------| |-------| + */ +void +xfs_iext_remove_indirect( + xfs_ifork_t *ifp, /* inode fork pointer */ + xfs_extnum_t idx, /* index to begin removing extents */ + int count) /* number of extents to remove */ +{ + xfs_ext_irec_t *erp; /* indirection array pointer */ + int erp_idx = 0; /* indirection array index */ + xfs_extnum_t ext_cnt; /* extents left to remove */ + xfs_extnum_t ext_diff; /* extents to remove in current list */ + xfs_extnum_t nex1; /* number of extents before idx */ + xfs_extnum_t nex2; /* extents after idx + count */ + int nlists; /* entries in indirecton array */ + int page_idx = idx; /* index in target extent list */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 0); + ASSERT(erp != NULL); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + nex1 = page_idx; + ext_cnt = count; + while (ext_cnt) { + nex2 = MAX((erp->er_extcount - (nex1 + ext_cnt)), 0); + ext_diff = MIN(ext_cnt, (erp->er_extcount - nex1)); + /* + * Check for deletion of entire list; + * xfs_iext_irec_remove() updates extent offsets. + */ + if (ext_diff == erp->er_extcount) { + xfs_iext_irec_remove(ifp, erp_idx); + ext_cnt -= ext_diff; + nex1 = 0; + if (ext_cnt) { + ASSERT(erp_idx < ifp->if_real_bytes / + XFS_IEXT_BUFSZ); + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + nex1 = 0; + continue; + } else { + break; + } + } + /* Move extents up (if needed) */ + if (nex2) { + memmove(&erp->er_extbuf[nex1], + &erp->er_extbuf[nex1 + ext_diff], + nex2 * sizeof(xfs_bmbt_rec_t)); + } + /* Zero out rest of page */ + memset(&erp->er_extbuf[nex1 + nex2], 0, (XFS_IEXT_BUFSZ - + ((nex1 + nex2) * sizeof(xfs_bmbt_rec_t)))); + /* Update remaining counters */ + erp->er_extcount -= ext_diff; + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -ext_diff); + ext_cnt -= ext_diff; + nex1 = 0; + erp_idx++; + erp++; + } + ifp->if_bytes -= count * sizeof(xfs_bmbt_rec_t); + xfs_iext_irec_compact(ifp); +} + /* * Create, destroy, or resize a linear (direct) block of extents. */ @@ -3827,6 +4088,10 @@ xfs_iext_realloc_direct( rnew_size = new_size; + ASSERT(!(ifp->if_flags & XFS_IFEXTIREC) || + ((new_size >= 0) && (new_size <= XFS_IEXT_BUFSZ) && + (new_size != ifp->if_real_bytes))); + /* Free extent records */ if (new_size == 0) { xfs_iext_destroy(ifp); @@ -3919,6 +4184,60 @@ xfs_iext_inline_to_direct( ifp->if_real_bytes = new_size; } +/* + * Resize an extent indirection array to new_size bytes. + */ +void +xfs_iext_realloc_indirect( + xfs_ifork_t *ifp, /* inode fork pointer */ + int new_size) /* new indirection array size */ +{ + int nlists; /* number of irec's (ex lists) */ + int size; /* current indirection array size */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + size = nlists * sizeof(xfs_ext_irec_t); + ASSERT(ifp->if_real_bytes); + ASSERT((new_size >= 0) && (new_size != size)); + if (new_size == 0) { + xfs_iext_destroy(ifp); + } else { + ifp->if_u1.if_ext_irec = (xfs_ext_irec_t *) + kmem_realloc(ifp->if_u1.if_ext_irec, + new_size, size, KM_SLEEP); + } +} + +/* + * Switch from indirection array to linear (direct) extent allocations. + */ +void +xfs_iext_indirect_to_direct( + xfs_ifork_t *ifp) /* inode fork pointer */ +{ + xfs_bmbt_rec_t *ep; /* extent record pointer */ + xfs_extnum_t nextents; /* number of extents in file */ + int size; /* size of file extents */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); + ASSERT(nextents <= XFS_LINEAR_EXTS); + size = nextents * sizeof(xfs_bmbt_rec_t); + + xfs_iext_irec_compact_full(ifp); + ASSERT(ifp->if_real_bytes == XFS_IEXT_BUFSZ); + + ep = ifp->if_u1.if_ext_irec->er_extbuf; + kmem_free(ifp->if_u1.if_ext_irec, sizeof(xfs_ext_irec_t)); + ifp->if_flags &= ~XFS_IFEXTIREC; + ifp->if_u1.if_extents = ep; + ifp->if_bytes = size; + if (nextents < XFS_LINEAR_EXTS) { + xfs_iext_realloc_direct(ifp, size); + } +} + /* * Free incore file extents. */ @@ -3926,7 +4245,16 @@ void xfs_iext_destroy( xfs_ifork_t *ifp) /* inode fork pointer */ { - if (ifp->if_real_bytes) { + if (ifp->if_flags & XFS_IFEXTIREC) { + int erp_idx; + int nlists; + + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + for (erp_idx = nlists - 1; erp_idx >= 0 ; erp_idx--) { + xfs_iext_irec_remove(ifp, erp_idx); + } + ifp->if_flags &= ~XFS_IFEXTIREC; + } else if (ifp->if_real_bytes) { kmem_free(ifp->if_u1.if_extents, ifp->if_real_bytes); } else if (ifp->if_bytes) { memset(ifp->if_u2.if_inline_ext, 0, XFS_INLINE_EXTS * @@ -3936,3 +4264,380 @@ xfs_iext_destroy( ifp->if_real_bytes = 0; ifp->if_bytes = 0; } + +/* + * Return a pointer to the indirection array entry containing the + * extent record for filesystem block bno. Store the index of the + * target irec in *erp_idxp. + */ +xfs_ext_irec_t * +xfs_iext_bno_to_irec( + xfs_ifork_t *ifp, /* inode fork pointer */ + xfs_fileoff_t bno, /* block number to search for */ + int *erp_idxp) /* irec index of target ext list */ +{ + xfs_ext_irec_t *erp = NULL; /* indirection array pointer */ + xfs_ext_irec_t *erp_next; /* next indirection array entry */ + xfs_extnum_t erp_idx; /* indirection array index */ + int nlists; /* number of extent irec's (lists) */ + int high; /* binary search upper limit */ + int low; /* binary search lower limit */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + erp_idx = 0; + low = 0; + high = nlists - 1; + while (low <= high) { + erp_idx = (low + high) >> 1; + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + erp_next = erp_idx < nlists - 1 ? erp + 1 : NULL; + if (bno < xfs_bmbt_get_startoff(erp->er_extbuf)) { + high = erp_idx - 1; + } else if (erp_next && bno >= + xfs_bmbt_get_startoff(erp_next->er_extbuf)) { + low = erp_idx + 1; + } else { + break; + } + } + *erp_idxp = erp_idx; + return erp; +} + +/* + * Return a pointer to the indirection array entry containing the + * extent record at file extent index *idxp. Store the index of the + * target irec in *erp_idxp and store the page index of the target + * extent record in *idxp. + */ +xfs_ext_irec_t * +xfs_iext_idx_to_irec( + xfs_ifork_t *ifp, /* inode fork pointer */ + xfs_extnum_t *idxp, /* extent index (file -> page) */ + int *erp_idxp, /* pointer to target irec */ + int realloc) /* new bytes were just added */ +{ + xfs_ext_irec_t *prev; /* pointer to previous irec */ + xfs_ext_irec_t *erp = NULL; /* pointer to current irec */ + int erp_idx; /* indirection array index */ + int nlists; /* number of irec's (ex lists) */ + int high; /* binary search upper limit */ + int low; /* binary search lower limit */ + xfs_extnum_t page_idx = *idxp; /* extent index in target list */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + ASSERT(page_idx >= 0 && page_idx <= + ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + erp_idx = 0; + low = 0; + high = nlists - 1; + + /* Binary search extent irec's */ + while (low <= high) { + erp_idx = (low + high) >> 1; + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + prev = erp_idx > 0 ? erp - 1 : NULL; + if (page_idx < erp->er_extoff || (page_idx == erp->er_extoff && + realloc && prev && prev->er_extcount < XFS_LINEAR_EXTS)) { + high = erp_idx - 1; + } else if (page_idx > erp->er_extoff + erp->er_extcount || + (page_idx == erp->er_extoff + erp->er_extcount && + !realloc)) { + low = erp_idx + 1; + } else if (page_idx == erp->er_extoff + erp->er_extcount && + erp->er_extcount == XFS_LINEAR_EXTS) { + ASSERT(realloc); + page_idx = 0; + erp_idx++; + erp = erp_idx < nlists ? erp + 1 : NULL; + break; + } else { + page_idx -= erp->er_extoff; + break; + } + } + *idxp = page_idx; + *erp_idxp = erp_idx; + return(erp); +} + +/* + * Allocate and initialize an indirection array once the space needed + * for incore extents increases above XFS_IEXT_BUFSZ. + */ +void +xfs_iext_irec_init( + xfs_ifork_t *ifp) /* inode fork pointer */ +{ + xfs_ext_irec_t *erp; /* indirection array pointer */ + xfs_extnum_t nextents; /* number of extents in file */ + + ASSERT(!(ifp->if_flags & XFS_IFEXTIREC)); + nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); + ASSERT(nextents <= XFS_LINEAR_EXTS); + + erp = (xfs_ext_irec_t *) + kmem_alloc(sizeof(xfs_ext_irec_t), KM_SLEEP); + + if (nextents == 0) { + ifp->if_u1.if_extents = (xfs_bmbt_rec_t *) + kmem_alloc(XFS_IEXT_BUFSZ, KM_SLEEP); + } else if (!ifp->if_real_bytes) { + xfs_iext_inline_to_direct(ifp, XFS_IEXT_BUFSZ); + } else if (ifp->if_real_bytes < XFS_IEXT_BUFSZ) { + xfs_iext_realloc_direct(ifp, XFS_IEXT_BUFSZ); + } + erp->er_extbuf = ifp->if_u1.if_extents; + erp->er_extcount = nextents; + erp->er_extoff = 0; + + ifp->if_flags |= XFS_IFEXTIREC; + ifp->if_real_bytes = XFS_IEXT_BUFSZ; + ifp->if_bytes = nextents * sizeof(xfs_bmbt_rec_t); + ifp->if_u1.if_ext_irec = erp; + + return; +} + +/* + * Allocate and initialize a new entry in the indirection array. + */ +xfs_ext_irec_t * +xfs_iext_irec_new( + xfs_ifork_t *ifp, /* inode fork pointer */ + int erp_idx) /* index for new irec */ +{ + xfs_ext_irec_t *erp; /* indirection array pointer */ + int i; /* loop counter */ + int nlists; /* number of irec's (ex lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + + /* Resize indirection array */ + xfs_iext_realloc_indirect(ifp, ++nlists * + sizeof(xfs_ext_irec_t)); + /* + * Move records down in the array so the + * new page can use erp_idx. + */ + erp = ifp->if_u1.if_ext_irec; + for (i = nlists - 1; i > erp_idx; i--) { + memmove(&erp[i], &erp[i-1], sizeof(xfs_ext_irec_t)); + } + ASSERT(i == erp_idx); + + /* Initialize new extent record */ + erp = ifp->if_u1.if_ext_irec; + erp[erp_idx].er_extbuf = (xfs_bmbt_rec_t *) + kmem_alloc(XFS_IEXT_BUFSZ, KM_SLEEP); + ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ; + memset(erp[erp_idx].er_extbuf, 0, XFS_IEXT_BUFSZ); + erp[erp_idx].er_extcount = 0; + erp[erp_idx].er_extoff = erp_idx > 0 ? + erp[erp_idx-1].er_extoff + erp[erp_idx-1].er_extcount : 0; + return (&erp[erp_idx]); +} + +/* + * Remove a record from the indirection array. + */ +void +xfs_iext_irec_remove( + xfs_ifork_t *ifp, /* inode fork pointer */ + int erp_idx) /* irec index to remove */ +{ + xfs_ext_irec_t *erp; /* indirection array pointer */ + int i; /* loop counter */ + int nlists; /* number of irec's (ex lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + if (erp->er_extbuf) { + xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, + -erp->er_extcount); + kmem_free(erp->er_extbuf, XFS_IEXT_BUFSZ); + } + /* Compact extent records */ + erp = ifp->if_u1.if_ext_irec; + for (i = erp_idx; i < nlists - 1; i++) { + memmove(&erp[i], &erp[i+1], sizeof(xfs_ext_irec_t)); + } + /* + * Manually free the last extent record from the indirection + * array. A call to xfs_iext_realloc_indirect() with a size + * of zero would result in a call to xfs_iext_destroy() which + * would in turn call this function again, creating a nasty + * infinite loop. + */ + if (--nlists) { + xfs_iext_realloc_indirect(ifp, + nlists * sizeof(xfs_ext_irec_t)); + } else { + kmem_free(ifp->if_u1.if_ext_irec, + sizeof(xfs_ext_irec_t)); + } + ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ; +} + +/* + * This is called to clean up large amounts of unused memory allocated + * by the indirection array. Before compacting anything though, verify + * that the indirection array is still needed and switch back to the + * linear extent list (or even the inline buffer) if possible. The + * compaction policy is as follows: + * + * Full Compaction: Extents fit into a single page (or inline buffer) + * Full Compaction: Extents occupy less than 10% of allocated space + * Partial Compaction: Extents occupy > 10% and < 50% of allocated space + * No Compaction: Extents occupy at least 50% of allocated space + */ +void +xfs_iext_irec_compact( + xfs_ifork_t *ifp) /* inode fork pointer */ +{ + xfs_extnum_t nextents; /* number of extents in file */ + int nlists; /* number of irec's (ex lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); + + if (nextents == 0) { + xfs_iext_destroy(ifp); + } else if (nextents <= XFS_INLINE_EXTS) { + xfs_iext_indirect_to_direct(ifp); + xfs_iext_direct_to_inline(ifp, nextents); + } else if (nextents <= XFS_LINEAR_EXTS) { + xfs_iext_indirect_to_direct(ifp); + } else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 3) { + xfs_iext_irec_compact_full(ifp); + } else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 1) { + xfs_iext_irec_compact_pages(ifp); + } +} + +/* + * Combine extents from neighboring extent pages. + */ +void +xfs_iext_irec_compact_pages( + xfs_ifork_t *ifp) /* inode fork pointer */ +{ + xfs_ext_irec_t *erp, *erp_next;/* pointers to irec entries */ + int erp_idx = 0; /* indirection array index */ + int nlists; /* number of irec's (ex lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + while (erp_idx < nlists - 1) { + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + erp_next = erp + 1; + if (erp_next->er_extcount <= + (XFS_LINEAR_EXTS - erp->er_extcount)) { + memmove(&erp->er_extbuf[erp->er_extcount], + erp_next->er_extbuf, erp_next->er_extcount * + sizeof(xfs_bmbt_rec_t)); + erp->er_extcount += erp_next->er_extcount; + /* + * Free page before removing extent record + * so er_extoffs don't get modified in + * xfs_iext_irec_remove. + */ + kmem_free(erp_next->er_extbuf, XFS_IEXT_BUFSZ); + erp_next->er_extbuf = NULL; + xfs_iext_irec_remove(ifp, erp_idx + 1); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + } else { + erp_idx++; + } + } +} + +/* + * Fully compact the extent records managed by the indirection array. + */ +void +xfs_iext_irec_compact_full( + xfs_ifork_t *ifp) /* inode fork pointer */ +{ + xfs_bmbt_rec_t *ep, *ep_next; /* extent record pointers */ + xfs_ext_irec_t *erp, *erp_next; /* extent irec pointers */ + int erp_idx = 0; /* extent irec index */ + int ext_avail; /* empty entries in ex list */ + int ext_diff; /* number of exts to add */ + int nlists; /* number of irec's (ex lists) */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + erp = ifp->if_u1.if_ext_irec; + ep = &erp->er_extbuf[erp->er_extcount]; + erp_next = erp + 1; + ep_next = erp_next->er_extbuf; + while (erp_idx < nlists - 1) { + ext_avail = XFS_LINEAR_EXTS - erp->er_extcount; + ext_diff = MIN(ext_avail, erp_next->er_extcount); + memcpy(ep, ep_next, ext_diff * sizeof(xfs_bmbt_rec_t)); + erp->er_extcount += ext_diff; + erp_next->er_extcount -= ext_diff; + /* Remove next page */ + if (erp_next->er_extcount == 0) { + /* + * Free page before removing extent record + * so er_extoffs don't get modified in + * xfs_iext_irec_remove. + */ + kmem_free(erp_next->er_extbuf, + erp_next->er_extcount * sizeof(xfs_bmbt_rec_t)); + erp_next->er_extbuf = NULL; + xfs_iext_irec_remove(ifp, erp_idx + 1); + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + /* Update next page */ + } else { + /* Move rest of page up to become next new page */ + memmove(erp_next->er_extbuf, ep_next, + erp_next->er_extcount * sizeof(xfs_bmbt_rec_t)); + ep_next = erp_next->er_extbuf; + memset(&ep_next[erp_next->er_extcount], 0, + (XFS_LINEAR_EXTS - erp_next->er_extcount) * + sizeof(xfs_bmbt_rec_t)); + } + if (erp->er_extcount == XFS_LINEAR_EXTS) { + erp_idx++; + if (erp_idx < nlists) + erp = &ifp->if_u1.if_ext_irec[erp_idx]; + else + break; + } + ep = &erp->er_extbuf[erp->er_extcount]; + erp_next = erp + 1; + ep_next = erp_next->er_extbuf; + } +} + +/* + * This is called to update the er_extoff field in the indirection + * array when extents have been added or removed from one of the + * extent lists. erp_idx contains the irec index to begin updating + * at and ext_diff contains the number of extents that were added + * or removed. + */ +void +xfs_iext_irec_update_extoffs( + xfs_ifork_t *ifp, /* inode fork pointer */ + int erp_idx, /* irec index to update */ + int ext_diff) /* number of new extents */ +{ + int i; /* loop counter */ + int nlists; /* number of irec's (ex lists */ + + ASSERT(ifp->if_flags & XFS_IFEXTIREC); + nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ; + for (i = erp_idx; i < nlists; i++) { + ifp->if_u1.if_ext_irec[i].er_extoff += ext_diff; + } +} diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h index 740b73fabd2..3c1df1d642f 100644 --- a/fs/xfs/xfs_inode.h +++ b/fs/xfs/xfs_inode.h @@ -24,11 +24,38 @@ #define XFS_DATA_FORK 0 #define XFS_ATTR_FORK 1 +/* + * The following xfs_ext_irec_t struct introduces a second (top) level + * to the in-core extent allocation scheme. These structs are allocated + * in a contiguous block, creating an indirection array where each entry + * (irec) contains a pointer to a buffer of in-core extent records which + * it manages. Each extent buffer is 4k in size, since 4k is the system + * page size on Linux i386 and systems with larger page sizes don't seem + * to gain much, if anything, by using their native page size as the + * extent buffer size. Also, using 4k extent buffers everywhere provides + * a consistent interface for CXFS across different platforms. + * + * There is currently no limit on the number of irec's (extent lists) + * allowed, so heavily fragmented files may require an indirection array + * which spans multiple system pages of memory. The number of extents + * which would require this amount of contiguous memory is very large + * and should not cause problems in the foreseeable future. However, + * if the memory needed for the contiguous array ever becomes a problem, + * it is possible that a third level of indirection may be required. + */ +typedef struct xfs_ext_irec { + xfs_bmbt_rec_t *er_extbuf; /* block of extent records */ + xfs_extnum_t er_extoff; /* extent offset in file */ + xfs_extnum_t er_extcount; /* number of extents in page/block */ +} xfs_ext_irec_t; + /* * File incore extent information, present for each of data & attr forks. */ -#define XFS_INLINE_EXTS 2 -#define XFS_INLINE_DATA 32 +#define XFS_IEXT_BUFSZ 4096 +#define XFS_LINEAR_EXTS (XFS_IEXT_BUFSZ / (uint)sizeof(xfs_bmbt_rec_t)) +#define XFS_INLINE_EXTS 2 +#define XFS_INLINE_DATA 32 typedef struct xfs_ifork { int if_bytes; /* bytes in if_u1 */ int if_real_bytes; /* bytes allocated in if_u1 */ @@ -39,6 +66,7 @@ typedef struct xfs_ifork { xfs_extnum_t if_lastex; /* last if_extents used */ union { xfs_bmbt_rec_t *if_extents; /* linear map file exts */ + xfs_ext_irec_t *if_ext_irec; /* irec map file exts */ char *if_data; /* inline file data */ } if_u1; union { @@ -61,9 +89,10 @@ typedef struct xfs_ifork { /* * Per-fork incore inode flags. */ -#define XFS_IFINLINE 0x0001 /* Inline data is read in */ -#define XFS_IFEXTENTS 0x0002 /* All extent pointers are read in */ -#define XFS_IFBROOT 0x0004 /* i_broot points to the bmap b-tree root */ +#define XFS_IFINLINE 0x01 /* Inline data is read in */ +#define XFS_IFEXTENTS 0x02 /* All extent pointers are read in */ +#define XFS_IFBROOT 0x04 /* i_broot points to the bmap b-tree root */ +#define XFS_IFEXTIREC 0x08 /* Indirection array of extent blocks */ /* * Flags for xfs_imap() and xfs_dilocate(). @@ -438,13 +467,26 @@ xfs_bmbt_rec_t *xfs_iext_get_ext(xfs_ifork_t *, xfs_extnum_t); void xfs_iext_insert(xfs_ifork_t *, xfs_extnum_t, xfs_extnum_t, xfs_bmbt_irec_t *); void xfs_iext_add(xfs_ifork_t *, xfs_extnum_t, int); +void xfs_iext_add_indirect_multi(xfs_ifork_t *, int, xfs_extnum_t, int); void xfs_iext_remove(xfs_ifork_t *, xfs_extnum_t, int); void xfs_iext_remove_inline(xfs_ifork_t *, xfs_extnum_t, int); void xfs_iext_remove_direct(xfs_ifork_t *, xfs_extnum_t, int); +void xfs_iext_remove_indirect(xfs_ifork_t *, xfs_extnum_t, int); void xfs_iext_realloc_direct(xfs_ifork_t *, int); +void xfs_iext_realloc_indirect(xfs_ifork_t *, int); +void xfs_iext_indirect_to_direct(xfs_ifork_t *); void xfs_iext_direct_to_inline(xfs_ifork_t *, xfs_extnum_t); void xfs_iext_inline_to_direct(xfs_ifork_t *, int); void xfs_iext_destroy(xfs_ifork_t *); +xfs_ext_irec_t *xfs_iext_bno_to_irec(xfs_ifork_t *, xfs_fileoff_t, int *); +xfs_ext_irec_t *xfs_iext_idx_to_irec(xfs_ifork_t *, xfs_extnum_t *, int *, int); +void xfs_iext_irec_init(xfs_ifork_t *); +xfs_ext_irec_t *xfs_iext_irec_new(xfs_ifork_t *, int); +void xfs_iext_irec_remove(xfs_ifork_t *, int); +void xfs_iext_irec_compact(xfs_ifork_t *); +void xfs_iext_irec_compact_pages(xfs_ifork_t *); +void xfs_iext_irec_compact_full(xfs_ifork_t *); +void xfs_iext_irec_update_extoffs(xfs_ifork_t *, int, int); #define xfs_ipincount(ip) ((unsigned int) atomic_read(&ip->i_pincount)) -- cgit v1.2.3-70-g09d2 From 39269e29d4aad04252e0debec4c9b01bac16a257 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 14 Mar 2006 13:33:50 +1100 Subject: [XFS] Reduce xfs_bmapi stack use by removing some local state variables, and directly testing flags instead. SGI-PV: 947312 SGI-Modid: xfs-linux-melb:xfs-kern:25370a Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 79 ++++++++++++++++++++++++++----------------------------- 1 file changed, 37 insertions(+), 42 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index 81a95b684b6..da8fa0cd79c 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -4629,10 +4629,6 @@ xfs_bmapi( xfs_btree_cur_t *cur; /* bmap btree cursor */ xfs_fileoff_t end; /* end of mapped file region */ int eof; /* we've hit the end of extents */ - char contig; /* allocation must be one extent */ - char delay; /* this request is for delayed alloc */ - char exact; /* don't do all of wasdelayed extent */ - char convert; /* unwritten extent I/O completion */ xfs_bmbt_rec_t *ep; /* extent record pointer */ int error; /* error return */ xfs_bmbt_irec_t got; /* current file extent record */ @@ -4651,13 +4647,9 @@ xfs_bmapi( int tmp_logflags; /* temp flags holder */ int whichfork; /* data or attr fork */ char inhole; /* current location is hole in file */ - char stateless; /* ignore state flag set */ - char trim; /* output trimmed to match range */ - char userdata; /* allocating non-metadata */ char wasdelay; /* old extent was delayed */ char wr; /* this is a write request */ char rt; /* this is a realtime file */ - char rsvd; /* OK to allocate reserved blocks */ #ifdef DEBUG xfs_fileoff_t orig_bno; /* original block number value */ int orig_flags; /* original flags arg value */ @@ -4694,15 +4686,8 @@ xfs_bmapi( XFS_STATS_INC(xs_blk_mapw); else XFS_STATS_INC(xs_blk_mapr); - delay = (flags & XFS_BMAPI_DELAY) != 0; - trim = (flags & XFS_BMAPI_ENTIRE) == 0; - userdata = (flags & XFS_BMAPI_METADATA) == 0; - convert = (flags & XFS_BMAPI_CONVERT) != 0; - exact = (flags & XFS_BMAPI_EXACT) != 0; - rsvd = (flags & XFS_BMAPI_RSVBLOCKS) != 0; - contig = (flags & XFS_BMAPI_CONTIG) != 0; /* - * stateless is used to combine extents which + * IGSTATE flag is used to combine extents which * differ only due to the state of the extents. * This technique is used from xfs_getbmap() * when the caller does not wish to see the @@ -4718,10 +4703,9 @@ xfs_bmapi( * xfs_strat_comp(), where the xfs_bmapi() call * is transactioned, and the extents combined. */ - stateless = (flags & XFS_BMAPI_IGSTATE) != 0; - if (stateless && wr) /* if writing unwritten space, no */ - wr = 0; /* allocations are allowed */ - ASSERT(wr || !delay); + if ((flags & XFS_BMAPI_IGSTATE) && wr) /* if writing unwritten space */ + wr = 0; /* no allocations are allowed */ + ASSERT(wr || !(flags & XFS_BMAPI_DELAY)); logflags = 0; nallocs = 0; cur = NULL; @@ -4756,7 +4740,7 @@ xfs_bmapi( if (eof && !wr) got.br_startoff = end; inhole = eof || got.br_startoff > bno; - wasdelay = wr && !inhole && !delay && + wasdelay = wr && !inhole && !(flags & XFS_BMAPI_DELAY) && ISNULLSTARTBLOCK(got.br_startblock); /* * First, deal with the hole before the allocated space @@ -4768,7 +4752,7 @@ xfs_bmapi( * allocate the stuff asked for in this bmap call * but that wouldn't be as good. */ - if (wasdelay && !exact) { + if (wasdelay && !(flags & XFS_BMAPI_EXACT)) { alen = (xfs_extlen_t)got.br_blockcount; aoff = got.br_startoff; if (lastx != NULLEXTNUM && lastx) { @@ -4790,8 +4774,8 @@ xfs_bmapi( got.br_startoff - bno); aoff = bno; } - minlen = contig ? alen : 1; - if (delay) { + minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1; + if (flags & XFS_BMAPI_DELAY) { xfs_extlen_t extsz; /* Figure out the extent size, adjust alen */ @@ -4804,7 +4788,9 @@ xfs_bmapi( if (extsz) { error = xfs_bmap_extsize_align(mp, &got, &prev, extsz, - rt, eof, delay, convert, + rt, eof, + flags&XFS_BMAPI_DELAY, + flags&XFS_BMAPI_CONVERT, &aoff, &alen); ASSERT(!error); } @@ -4842,24 +4828,29 @@ xfs_bmapi( if (rt) { error = xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS, - -(extsz), rsvd); + -(extsz), (flags & + XFS_BMAPI_RSVBLOCKS)); } else { error = xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, - -(alen), rsvd); + -(alen), (flags & + XFS_BMAPI_RSVBLOCKS)); } if (!error) { error = xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, - -(indlen), rsvd); + -(indlen), (flags & + XFS_BMAPI_RSVBLOCKS)); if (error && rt) xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS, - extsz, rsvd); + extsz, (flags & + XFS_BMAPI_RSVBLOCKS)); else if (error) xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, - alen, rsvd); + alen, (flags & + XFS_BMAPI_RSVBLOCKS)); } if (error) { @@ -4892,7 +4883,7 @@ xfs_bmapi( /* Indicate if this is the first user data * in the file, or just any user data. */ - if (userdata) { + if (!(flags & XFS_BMAPI_METADATA)) { bma.userdata = (aoff == 0) ? XFS_ALLOC_INITIAL_USER_DATA : XFS_ALLOC_USERDATA; @@ -4904,7 +4895,7 @@ xfs_bmapi( bma.firstblock = *firstblock; bma.alen = alen; bma.off = aoff; - bma.conv = convert; + bma.conv = (flags & XFS_BMAPI_CONVERT); bma.wasdel = wasdelay; bma.minlen = minlen; bma.low = flist->xbf_low; @@ -4915,7 +4906,8 @@ xfs_bmapi( * is larger than a stripe unit. */ if (mp->m_dalign && alen >= mp->m_dalign && - userdata && whichfork == XFS_DATA_FORK) { + (!(flags & XFS_BMAPI_METADATA)) && + (whichfork == XFS_DATA_FORK)) { if ((error = xfs_bmap_isaeof(ip, aoff, whichfork, &bma.aeof))) goto error0; @@ -4978,7 +4970,7 @@ xfs_bmapi( } error = xfs_bmap_add_extent(ip, lastx, &cur, &got, firstblock, flist, &tmp_logflags, whichfork, - rsvd); + (flags & XFS_BMAPI_RSVBLOCKS)); logflags |= tmp_logflags; if (error) goto error0; @@ -4990,7 +4982,7 @@ xfs_bmapi( ASSERT(got.br_startoff + got.br_blockcount >= aoff + alen); #ifdef DEBUG - if (delay) { + if (flags & XFS_BMAPI_DELAY) { ASSERT(ISNULLSTARTBLOCK(got.br_startblock)); ASSERT(STARTBLOCKVAL(got.br_startblock) > 0); } @@ -5019,14 +5011,15 @@ xfs_bmapi( * Then deal with the allocated space we found. */ ASSERT(ep != NULL); - if (trim && (got.br_startoff + got.br_blockcount > obno)) { + if (!(flags & XFS_BMAPI_ENTIRE) && + (got.br_startoff + got.br_blockcount > obno)) { if (obno > bno) bno = obno; ASSERT((bno >= obno) || (n == 0)); ASSERT(bno < end); mval->br_startoff = bno; if (ISNULLSTARTBLOCK(got.br_startblock)) { - ASSERT(!wr || delay); + ASSERT(!wr || (flags & XFS_BMAPI_DELAY)); mval->br_startblock = DELAYSTARTBLOCK; } else mval->br_startblock = @@ -5048,7 +5041,7 @@ xfs_bmapi( } else { *mval = got; if (ISNULLSTARTBLOCK(mval->br_startblock)) { - ASSERT(!wr || delay); + ASSERT(!wr || (flags & XFS_BMAPI_DELAY)); mval->br_startblock = DELAYSTARTBLOCK; } } @@ -5074,7 +5067,7 @@ xfs_bmapi( mval->br_state = XFS_EXT_NORM; error = xfs_bmap_add_extent(ip, lastx, &cur, mval, firstblock, flist, &tmp_logflags, whichfork, - rsvd); + (flags & XFS_BMAPI_RSVBLOCKS)); logflags |= tmp_logflags; if (error) goto error0; @@ -5091,9 +5084,10 @@ xfs_bmapi( continue; } - ASSERT(!trim || + ASSERT((flags & XFS_BMAPI_ENTIRE) || ((mval->br_startoff + mval->br_blockcount) <= end)); - ASSERT(!trim || (mval->br_blockcount <= len) || + ASSERT((flags & XFS_BMAPI_ENTIRE) || + (mval->br_blockcount <= len) || (mval->br_startoff < obno)); bno = mval->br_startoff + mval->br_blockcount; len = end - bno; @@ -5108,7 +5102,8 @@ xfs_bmapi( mval[-1].br_startblock != HOLESTARTBLOCK && mval->br_startblock == mval[-1].br_startblock + mval[-1].br_blockcount && - (stateless || mval[-1].br_state == mval->br_state)) { + ((flags & XFS_BMAPI_IGSTATE) || + mval[-1].br_state == mval->br_state)) { ASSERT(mval->br_startoff == mval[-1].br_startoff + mval[-1].br_blockcount); mval[-1].br_blockcount += mval->br_blockcount; -- cgit v1.2.3-70-g09d2 From a365bdd5e8fae9c592b9e4851d931016f9fdd868 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 14 Mar 2006 13:34:16 +1100 Subject: [XFS] Reduce stack usage within xfs_bmapi by rearranging some code, splitting realtime/btree allocators apart. Based on Glens original patches. SGI-PV: 947312 SGI-Modid: xfs-linux-melb:xfs-kern:25372a Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 668 ++++++++++++++++++++++++++++-------------------------- 1 file changed, 345 insertions(+), 323 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index da8fa0cd79c..9f0ed4869d4 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -2294,25 +2294,15 @@ xfs_bmap_extsize_align( #define XFS_ALLOC_GAP_UNITS 4 -/* - * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file. - * It figures out where to ask the underlying allocator to put the new extent. - */ STATIC int -xfs_bmap_alloc( +xfs_bmap_adjacent( xfs_bmalloca_t *ap) /* bmap alloc argument struct */ { xfs_fsblock_t adjust; /* adjustment to block numbers */ - xfs_alloctype_t atype=0; /* type for allocation routines */ - int error; /* error return value */ xfs_agnumber_t fb_agno; /* ag number of ap->firstblock */ xfs_mount_t *mp; /* mount point structure */ int nullfb; /* true if ap->firstblock isn't set */ int rt; /* true if inode is realtime */ - xfs_extlen_t prod = 0; /* product factor for allocators */ - xfs_extlen_t ralen = 0; /* realtime allocation length */ - xfs_extlen_t align; /* minimum allocation alignment */ - xfs_rtblock_t rtx; #define ISVALID(x,y) \ (rt ? \ @@ -2321,75 +2311,10 @@ xfs_bmap_alloc( XFS_FSB_TO_AGNO(mp, x) < mp->m_sb.sb_agcount && \ XFS_FSB_TO_AGBNO(mp, x) < mp->m_sb.sb_agblocks) - /* - * Set up variables. - */ mp = ap->ip->i_mount; nullfb = ap->firstblock == NULLFSBLOCK; rt = XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata; fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock); - if (rt) { - align = ap->ip->i_d.di_extsize ? - ap->ip->i_d.di_extsize : mp->m_sb.sb_rextsize; - /* Set prod to match the extent size */ - prod = align / mp->m_sb.sb_rextsize; - - error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp, - align, rt, ap->eof, 0, - ap->conv, &ap->off, &ap->alen); - if (error) - return error; - ASSERT(ap->alen); - ASSERT(ap->alen % mp->m_sb.sb_rextsize == 0); - - /* - * If the offset & length are not perfectly aligned - * then kill prod, it will just get us in trouble. - */ - if (do_mod(ap->off, align) || ap->alen % align) - prod = 1; - /* - * Set ralen to be the actual requested length in rtextents. - */ - ralen = ap->alen / mp->m_sb.sb_rextsize; - /* - * If the old value was close enough to MAXEXTLEN that - * we rounded up to it, cut it back so it's valid again. - * Note that if it's a really large request (bigger than - * MAXEXTLEN), we don't hear about that number, and can't - * adjust the starting point to match it. - */ - if (ralen * mp->m_sb.sb_rextsize >= MAXEXTLEN) - ralen = MAXEXTLEN / mp->m_sb.sb_rextsize; - /* - * If it's an allocation to an empty file at offset 0, - * pick an extent that will space things out in the rt area. - */ - if (ap->eof && ap->off == 0) { - error = xfs_rtpick_extent(mp, ap->tp, ralen, &rtx); - if (error) - return error; - ap->rval = rtx * mp->m_sb.sb_rextsize; - } else - ap->rval = 0; - } else { - align = (ap->userdata && ap->ip->i_d.di_extsize && - (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)) ? - ap->ip->i_d.di_extsize : 0; - if (unlikely(align)) { - error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp, - align, rt, - ap->eof, 0, ap->conv, - &ap->off, &ap->alen); - ASSERT(!error); - ASSERT(ap->alen); - } - if (nullfb) - ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino); - else - ap->rval = ap->firstblock; - } - /* * If allocating at eof, and there's a previous real block, * try to use it's last block as our starting point. @@ -2514,281 +2439,378 @@ xfs_bmap_alloc( else if (gotbno != NULLFSBLOCK) ap->rval = gotbno; } +#undef ISVALID + return 0; +} + +STATIC int +xfs_bmap_rtalloc( + xfs_bmalloca_t *ap) /* bmap alloc argument struct */ +{ + xfs_alloctype_t atype = 0; /* type for allocation routines */ + int error; /* error return value */ + xfs_mount_t *mp; /* mount point structure */ + xfs_extlen_t prod = 0; /* product factor for allocators */ + xfs_extlen_t ralen = 0; /* realtime allocation length */ + xfs_extlen_t align; /* minimum allocation alignment */ + xfs_rtblock_t rtx; /* realtime extent number */ + xfs_rtblock_t rtb; + + mp = ap->ip->i_mount; + align = ap->ip->i_d.di_extsize ? + ap->ip->i_d.di_extsize : mp->m_sb.sb_rextsize; + prod = align / mp->m_sb.sb_rextsize; + error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp, + align, 1, ap->eof, 0, + ap->conv, &ap->off, &ap->alen); + if (error) + return error; + ASSERT(ap->alen); + ASSERT(ap->alen % mp->m_sb.sb_rextsize == 0); + + /* + * If the offset & length are not perfectly aligned + * then kill prod, it will just get us in trouble. + */ + if (do_mod(ap->off, align) || ap->alen % align) + prod = 1; + /* + * Set ralen to be the actual requested length in rtextents. + */ + ralen = ap->alen / mp->m_sb.sb_rextsize; + /* + * If the old value was close enough to MAXEXTLEN that + * we rounded up to it, cut it back so it's valid again. + * Note that if it's a really large request (bigger than + * MAXEXTLEN), we don't hear about that number, and can't + * adjust the starting point to match it. + */ + if (ralen * mp->m_sb.sb_rextsize >= MAXEXTLEN) + ralen = MAXEXTLEN / mp->m_sb.sb_rextsize; + /* + * If it's an allocation to an empty file at offset 0, + * pick an extent that will space things out in the rt area. + */ + if (ap->eof && ap->off == 0) { + error = xfs_rtpick_extent(mp, ap->tp, ralen, &rtx); + if (error) + return error; + ap->rval = rtx * mp->m_sb.sb_rextsize; + } else { + ap->rval = 0; + } + + xfs_bmap_adjacent(ap); + + /* + * Realtime allocation, done through xfs_rtallocate_extent. + */ + atype = ap->rval == 0 ? XFS_ALLOCTYPE_ANY_AG : XFS_ALLOCTYPE_NEAR_BNO; + do_div(ap->rval, mp->m_sb.sb_rextsize); + rtb = ap->rval; + ap->alen = ralen; + if ((error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, ap->alen, + &ralen, atype, ap->wasdel, prod, &rtb))) + return error; + if (rtb == NULLFSBLOCK && prod > 1 && + (error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, + ap->alen, &ralen, atype, + ap->wasdel, 1, &rtb))) + return error; + ap->rval = rtb; + if (ap->rval != NULLFSBLOCK) { + ap->rval *= mp->m_sb.sb_rextsize; + ralen *= mp->m_sb.sb_rextsize; + ap->alen = ralen; + ap->ip->i_d.di_nblocks += ralen; + xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE); + if (ap->wasdel) + ap->ip->i_delayed_blks -= ralen; + /* + * Adjust the disk quota also. This was reserved + * earlier. + */ + XFS_TRANS_MOD_DQUOT_BYINO(mp, ap->tp, ap->ip, + ap->wasdel ? XFS_TRANS_DQ_DELRTBCOUNT : + XFS_TRANS_DQ_RTBCOUNT, (long) ralen); + } else { + ap->alen = 0; + } + return 0; +} + +STATIC int +xfs_bmap_btalloc( + xfs_bmalloca_t *ap) /* bmap alloc argument struct */ +{ + xfs_mount_t *mp; /* mount point structure */ + xfs_alloctype_t atype = 0; /* type for allocation routines */ + xfs_extlen_t align; /* minimum allocation alignment */ + xfs_agnumber_t ag; + xfs_agnumber_t fb_agno; /* ag number of ap->firstblock */ + xfs_agnumber_t startag; + xfs_alloc_arg_t args; + xfs_extlen_t blen; + xfs_extlen_t delta; + xfs_extlen_t longest; + xfs_extlen_t need; + xfs_extlen_t nextminlen = 0; + xfs_perag_t *pag; + int nullfb; /* true if ap->firstblock isn't set */ + int isaligned; + int notinit; + int tryagain; + int error; + + mp = ap->ip->i_mount; + align = (ap->userdata && ap->ip->i_d.di_extsize && + (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)) ? + ap->ip->i_d.di_extsize : 0; + if (unlikely(align)) { + error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp, + align, 0, ap->eof, 0, ap->conv, + &ap->off, &ap->alen); + ASSERT(!error); + ASSERT(ap->alen); + } + nullfb = ap->firstblock == NULLFSBLOCK; + fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock); + if (nullfb) + ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino); + else + ap->rval = ap->firstblock; + + xfs_bmap_adjacent(ap); + /* * If allowed, use ap->rval; otherwise must use firstblock since * it's in the right allocation group. */ - if (nullfb || rt || XFS_FSB_TO_AGNO(mp, ap->rval) == fb_agno) + if (nullfb || XFS_FSB_TO_AGNO(mp, ap->rval) == fb_agno) ; else ap->rval = ap->firstblock; /* - * Realtime allocation, done through xfs_rtallocate_extent. + * Normal allocation, done through xfs_alloc_vextent. */ - if (rt) { -#ifndef __KERNEL__ - ASSERT(0); -#else - xfs_rtblock_t rtb; - - atype = ap->rval == 0 ? - XFS_ALLOCTYPE_ANY_AG : XFS_ALLOCTYPE_NEAR_BNO; - do_div(ap->rval, mp->m_sb.sb_rextsize); - rtb = ap->rval; - ap->alen = ralen; - if ((error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, ap->alen, - &ralen, atype, ap->wasdel, prod, &rtb))) - return error; - if (rtb == NULLFSBLOCK && prod > 1 && - (error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, - ap->alen, &ralen, atype, - ap->wasdel, 1, &rtb))) - return error; - ap->rval = rtb; - if (ap->rval != NULLFSBLOCK) { - ap->rval *= mp->m_sb.sb_rextsize; - ralen *= mp->m_sb.sb_rextsize; - ap->alen = ralen; - ap->ip->i_d.di_nblocks += ralen; - xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE); - if (ap->wasdel) - ap->ip->i_delayed_blks -= ralen; + tryagain = isaligned = 0; + args.tp = ap->tp; + args.mp = mp; + args.fsbno = ap->rval; + args.maxlen = MIN(ap->alen, mp->m_sb.sb_agblocks); + blen = 0; + if (nullfb) { + args.type = XFS_ALLOCTYPE_START_BNO; + args.total = ap->total; + /* + * Find the longest available space. + * We're going to try for the whole allocation at once. + */ + startag = ag = XFS_FSB_TO_AGNO(mp, args.fsbno); + notinit = 0; + down_read(&mp->m_peraglock); + while (blen < ap->alen) { + pag = &mp->m_perag[ag]; + if (!pag->pagf_init && + (error = xfs_alloc_pagf_init(mp, args.tp, + ag, XFS_ALLOC_FLAG_TRYLOCK))) { + up_read(&mp->m_peraglock); + return error; + } /* - * Adjust the disk quota also. This was reserved - * earlier. + * See xfs_alloc_fix_freelist... */ - XFS_TRANS_MOD_DQUOT_BYINO(mp, ap->tp, ap->ip, - ap->wasdel ? XFS_TRANS_DQ_DELRTBCOUNT : - XFS_TRANS_DQ_RTBCOUNT, - (long) ralen); - } else - ap->alen = 0; -#endif /* __KERNEL__ */ + if (pag->pagf_init) { + need = XFS_MIN_FREELIST_PAG(pag, mp); + delta = need > pag->pagf_flcount ? + need - pag->pagf_flcount : 0; + longest = (pag->pagf_longest > delta) ? + (pag->pagf_longest - delta) : + (pag->pagf_flcount > 0 || + pag->pagf_longest > 0); + if (blen < longest) + blen = longest; + } else + notinit = 1; + if (++ag == mp->m_sb.sb_agcount) + ag = 0; + if (ag == startag) + break; + } + up_read(&mp->m_peraglock); + /* + * Since the above loop did a BUF_TRYLOCK, it is + * possible that there is space for this request. + */ + if (notinit || blen < ap->minlen) + args.minlen = ap->minlen; + /* + * If the best seen length is less than the request + * length, use the best as the minimum. + */ + else if (blen < ap->alen) + args.minlen = blen; + /* + * Otherwise we've seen an extent as big as alen, + * use that as the minimum. + */ + else + args.minlen = ap->alen; + } else if (ap->low) { + args.type = XFS_ALLOCTYPE_FIRST_AG; + args.total = args.minlen = ap->minlen; + } else { + args.type = XFS_ALLOCTYPE_NEAR_BNO; + args.total = ap->total; + args.minlen = ap->minlen; + } + if (unlikely(ap->userdata && ap->ip->i_d.di_extsize && + (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE))) { + args.prod = ap->ip->i_d.di_extsize; + if ((args.mod = (xfs_extlen_t)do_mod(ap->off, args.prod))) + args.mod = (xfs_extlen_t)(args.prod - args.mod); + } else if (unlikely(mp->m_sb.sb_blocksize >= NBPP)) { + args.prod = 1; + args.mod = 0; + } else { + args.prod = NBPP >> mp->m_sb.sb_blocklog; + if ((args.mod = (xfs_extlen_t)(do_mod(ap->off, args.prod)))) + args.mod = (xfs_extlen_t)(args.prod - args.mod); } /* - * Normal allocation, done through xfs_alloc_vextent. + * If we are not low on available data blocks, and the + * underlying logical volume manager is a stripe, and + * the file offset is zero then try to allocate data + * blocks on stripe unit boundary. + * NOTE: ap->aeof is only set if the allocation length + * is >= the stripe unit and the allocation offset is + * at the end of file. */ - else { - xfs_agnumber_t ag; - xfs_alloc_arg_t args; - xfs_extlen_t blen; - xfs_extlen_t delta; - int isaligned; - xfs_extlen_t longest; - xfs_extlen_t need; - xfs_extlen_t nextminlen=0; - int notinit; - xfs_perag_t *pag; - xfs_agnumber_t startag; - int tryagain; - - tryagain = isaligned = 0; - args.tp = ap->tp; - args.mp = mp; - args.fsbno = ap->rval; - args.maxlen = MIN(ap->alen, mp->m_sb.sb_agblocks); - blen = 0; - if (nullfb) { - args.type = XFS_ALLOCTYPE_START_BNO; - args.total = ap->total; - /* - * Find the longest available space. - * We're going to try for the whole allocation at once. - */ - startag = ag = XFS_FSB_TO_AGNO(mp, args.fsbno); - notinit = 0; - down_read(&mp->m_peraglock); - while (blen < ap->alen) { - pag = &mp->m_perag[ag]; - if (!pag->pagf_init && - (error = xfs_alloc_pagf_init(mp, args.tp, - ag, XFS_ALLOC_FLAG_TRYLOCK))) { - up_read(&mp->m_peraglock); - return error; - } - /* - * See xfs_alloc_fix_freelist... - */ - if (pag->pagf_init) { - need = XFS_MIN_FREELIST_PAG(pag, mp); - delta = need > pag->pagf_flcount ? - need - pag->pagf_flcount : 0; - longest = (pag->pagf_longest > delta) ? - (pag->pagf_longest - delta) : - (pag->pagf_flcount > 0 || - pag->pagf_longest > 0); - if (blen < longest) - blen = longest; - } else - notinit = 1; - if (++ag == mp->m_sb.sb_agcount) - ag = 0; - if (ag == startag) - break; - } - up_read(&mp->m_peraglock); + if (!ap->low && ap->aeof) { + if (!ap->off) { + args.alignment = mp->m_dalign; + atype = args.type; + isaligned = 1; /* - * Since the above loop did a BUF_TRYLOCK, it is - * possible that there is space for this request. + * Adjust for alignment */ - if (notinit || blen < ap->minlen) - args.minlen = ap->minlen; + if (blen > args.alignment && blen <= ap->alen) + args.minlen = blen - args.alignment; + args.minalignslop = 0; + } else { /* - * If the best seen length is less than the request - * length, use the best as the minimum. + * First try an exact bno allocation. + * If it fails then do a near or start bno + * allocation with alignment turned on. */ - else if (blen < ap->alen) - args.minlen = blen; + atype = args.type; + tryagain = 1; + args.type = XFS_ALLOCTYPE_THIS_BNO; + args.alignment = 1; /* - * Otherwise we've seen an extent as big as alen, - * use that as the minimum. + * Compute the minlen+alignment for the + * next case. Set slop so that the value + * of minlen+alignment+slop doesn't go up + * between the calls. */ + if (blen > mp->m_dalign && blen <= ap->alen) + nextminlen = blen - mp->m_dalign; else - args.minlen = ap->alen; - } else if (ap->low) { - args.type = XFS_ALLOCTYPE_FIRST_AG; - args.total = args.minlen = ap->minlen; - } else { - args.type = XFS_ALLOCTYPE_NEAR_BNO; - args.total = ap->total; - args.minlen = ap->minlen; - } - if (unlikely(ap->userdata && ap->ip->i_d.di_extsize && - (ap->ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE))) { - args.prod = ap->ip->i_d.di_extsize; - if ((args.mod = (xfs_extlen_t)do_mod(ap->off, args.prod))) - args.mod = (xfs_extlen_t)(args.prod - args.mod); - } else if (unlikely(mp->m_sb.sb_blocksize >= NBPP)) { - args.prod = 1; - args.mod = 0; - } else { - args.prod = NBPP >> mp->m_sb.sb_blocklog; - if ((args.mod = (xfs_extlen_t)(do_mod(ap->off, args.prod)))) - args.mod = (xfs_extlen_t)(args.prod - args.mod); + nextminlen = args.minlen; + if (nextminlen + mp->m_dalign > args.minlen + 1) + args.minalignslop = + nextminlen + mp->m_dalign - + args.minlen - 1; + else + args.minalignslop = 0; } + } else { + args.alignment = 1; + args.minalignslop = 0; + } + args.minleft = ap->minleft; + args.wasdel = ap->wasdel; + args.isfl = 0; + args.userdata = ap->userdata; + if ((error = xfs_alloc_vextent(&args))) + return error; + if (tryagain && args.fsbno == NULLFSBLOCK) { /* - * If we are not low on available data blocks, and the - * underlying logical volume manager is a stripe, and - * the file offset is zero then try to allocate data - * blocks on stripe unit boundary. - * NOTE: ap->aeof is only set if the allocation length - * is >= the stripe unit and the allocation offset is - * at the end of file. + * Exact allocation failed. Now try with alignment + * turned on. */ - if (!ap->low && ap->aeof) { - if (!ap->off) { - args.alignment = mp->m_dalign; - atype = args.type; - isaligned = 1; - /* - * Adjust for alignment - */ - if (blen > args.alignment && blen <= ap->alen) - args.minlen = blen - args.alignment; - args.minalignslop = 0; - } else { - /* - * First try an exact bno allocation. - * If it fails then do a near or start bno - * allocation with alignment turned on. - */ - atype = args.type; - tryagain = 1; - args.type = XFS_ALLOCTYPE_THIS_BNO; - args.alignment = 1; - /* - * Compute the minlen+alignment for the - * next case. Set slop so that the value - * of minlen+alignment+slop doesn't go up - * between the calls. - */ - if (blen > mp->m_dalign && blen <= ap->alen) - nextminlen = blen - mp->m_dalign; - else - nextminlen = args.minlen; - if (nextminlen + mp->m_dalign > args.minlen + 1) - args.minalignslop = - nextminlen + mp->m_dalign - - args.minlen - 1; - else - args.minalignslop = 0; - } - } else { - args.alignment = 1; - args.minalignslop = 0; - } - args.minleft = ap->minleft; - args.wasdel = ap->wasdel; - args.isfl = 0; - args.userdata = ap->userdata; + args.type = atype; + args.fsbno = ap->rval; + args.alignment = mp->m_dalign; + args.minlen = nextminlen; + args.minalignslop = 0; + isaligned = 1; + if ((error = xfs_alloc_vextent(&args))) + return error; + } + if (isaligned && args.fsbno == NULLFSBLOCK) { + /* + * allocation failed, so turn off alignment and + * try again. + */ + args.type = atype; + args.fsbno = ap->rval; + args.alignment = 0; + if ((error = xfs_alloc_vextent(&args))) + return error; + } + if (args.fsbno == NULLFSBLOCK && nullfb && + args.minlen > ap->minlen) { + args.minlen = ap->minlen; + args.type = XFS_ALLOCTYPE_START_BNO; + args.fsbno = ap->rval; if ((error = xfs_alloc_vextent(&args))) return error; - if (tryagain && args.fsbno == NULLFSBLOCK) { - /* - * Exact allocation failed. Now try with alignment - * turned on. - */ - args.type = atype; - args.fsbno = ap->rval; - args.alignment = mp->m_dalign; - args.minlen = nextminlen; - args.minalignslop = 0; - isaligned = 1; - if ((error = xfs_alloc_vextent(&args))) - return error; - } - if (isaligned && args.fsbno == NULLFSBLOCK) { - /* - * allocation failed, so turn off alignment and - * try again. - */ - args.type = atype; - args.fsbno = ap->rval; - args.alignment = 0; - if ((error = xfs_alloc_vextent(&args))) - return error; - } - if (args.fsbno == NULLFSBLOCK && nullfb && - args.minlen > ap->minlen) { - args.minlen = ap->minlen; - args.type = XFS_ALLOCTYPE_START_BNO; - args.fsbno = ap->rval; - if ((error = xfs_alloc_vextent(&args))) - return error; - } - if (args.fsbno == NULLFSBLOCK && nullfb) { - args.fsbno = 0; - args.type = XFS_ALLOCTYPE_FIRST_AG; - args.total = ap->minlen; - args.minleft = 0; - if ((error = xfs_alloc_vextent(&args))) - return error; - ap->low = 1; - } - if (args.fsbno != NULLFSBLOCK) { - ap->firstblock = ap->rval = args.fsbno; - ASSERT(nullfb || fb_agno == args.agno || - (ap->low && fb_agno < args.agno)); - ap->alen = args.len; - ap->ip->i_d.di_nblocks += args.len; - xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE); - if (ap->wasdel) - ap->ip->i_delayed_blks -= args.len; - /* - * Adjust the disk quota also. This was reserved - * earlier. - */ - XFS_TRANS_MOD_DQUOT_BYINO(mp, ap->tp, ap->ip, - ap->wasdel ? XFS_TRANS_DQ_DELBCOUNT : - XFS_TRANS_DQ_BCOUNT, - (long) args.len); - } else { - ap->rval = NULLFSBLOCK; - ap->alen = 0; - } + } + if (args.fsbno == NULLFSBLOCK && nullfb) { + args.fsbno = 0; + args.type = XFS_ALLOCTYPE_FIRST_AG; + args.total = ap->minlen; + args.minleft = 0; + if ((error = xfs_alloc_vextent(&args))) + return error; + ap->low = 1; + } + if (args.fsbno != NULLFSBLOCK) { + ap->firstblock = ap->rval = args.fsbno; + ASSERT(nullfb || fb_agno == args.agno || + (ap->low && fb_agno < args.agno)); + ap->alen = args.len; + ap->ip->i_d.di_nblocks += args.len; + xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE); + if (ap->wasdel) + ap->ip->i_delayed_blks -= args.len; + /* + * Adjust the disk quota also. This was reserved + * earlier. + */ + XFS_TRANS_MOD_DQUOT_BYINO(mp, ap->tp, ap->ip, + ap->wasdel ? XFS_TRANS_DQ_DELBCOUNT : + XFS_TRANS_DQ_BCOUNT, + (long) args.len); + } else { + ap->rval = NULLFSBLOCK; + ap->alen = 0; } return 0; -#undef ISVALID +} + +/* + * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file. + * It figures out where to ask the underlying allocator to put the new extent. + */ +STATIC int +xfs_bmap_alloc( + xfs_bmalloca_t *ap) /* bmap alloc argument struct */ +{ + if ((ap->ip->i_d.di_flags & XFS_DIFLAG_REALTIME) && ap->userdata) + return xfs_bmap_rtalloc(ap); + return xfs_bmap_btalloc(ap); } /* -- cgit v1.2.3-70-g09d2 From f020b67f3c0c071ed5b2281bfa717125c8a39ad6 Mon Sep 17 00:00:00 2001 From: Mandy Kirkconnell Date: Tue, 14 Mar 2006 14:07:24 +1100 Subject: [XFS] Fix assert to check that in-core extents are inline only. SGI-PV: 950678 SGI-Modid: xfs-linux-melb:xfs-kern:207634a Signed-off-by: Mandy Kirkconnell Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index 9f0ed4869d4..64a02eaf1df 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -3414,7 +3414,8 @@ xfs_bmap_local_to_extents( args.tp = tp; args.mp = ip->i_mount; - ASSERT(ifp->if_flags & XFS_IFINLINE); + ASSERT((ifp->if_flags & + (XFS_IFINLINE|XFS_IFEXTENTS|XFS_IFEXTIREC)) == XFS_IFINLINE); /* * Allocate a block. We know we need only one, since the * file currently fits in an inode. @@ -3445,7 +3446,6 @@ xfs_bmap_local_to_extents( xfs_bmap_forkoff_reset(args.mp, ip, whichfork); xfs_idata_realloc(ip, -ifp->if_bytes, whichfork); xfs_iext_add(ifp, 0, 1); - ASSERT((ifp->if_flags & (XFS_IFEXTENTS|XFS_IFEXTIREC)) == XFS_IFEXTENTS); ep = xfs_iext_get_ext(ifp, 0); xfs_bmbt_set_allf(ep, 0, args.fsbno, 1, XFS_EXT_NORM); xfs_bmap_trace_post_update(fname, "new", ip, 0, whichfork); -- cgit v1.2.3-70-g09d2 From 8867bc9bf0aed7181aa72c7c938c6ce830b75166 Mon Sep 17 00:00:00 2001 From: Mandy Kirkconnell Date: Fri, 17 Mar 2006 17:25:04 +1100 Subject: [XFS] There are a few problems with the new xfs_bmap_search_multi_extents() wrapper function that I introduced in mod xfs-linux:xfs-kern:207393a. The function was added as a wrapper around xfs_bmap_do_search_extents() to avoid breaking the top-of-tree CXFS interface. The idea of the function was basically to extract the target extent buffer (if muli- level extent allocation mode), then call xfs_bmap_do_search_extents() with either a pointer to the first extent in the target buffer or a pointer to the first extent in the file, depending on which extent mode was being used. However, in addition to locating the target extent record for block bno, xfs_bmap_do_search_extents() also sets four parameters needed by the caller: *lastx, *eofp, *gotp, *prevp. Passing only the target extent buffer to xfs_bmap_do_search_extents() causes *eofp to be set incorrectly if the extent is at the end of the target list but there are actually more extents in the next er_extbuf. Likewise, if the extent is the first one in the buffer but NOT the first in the file, *prevp is incorrectly set to NULL. Adding the needed functionality to xfs_bmap_search_multi_extents() to re-set any incorrectly set fields is redundant and makes the call to xfs_bmap_do_search_extents() not make much sense when multi-level extent allocation mode is being used. This mod basically extracts the two functional components from xfs_bmap_do_search_extents(), with the intent of obsoleting/removing xfs_bmap_do_search_extents() after the CXFS mult-level in-core extent changes are checked in. The two components are: 1) The binary search to locate the target extent record, and 2) Setting the four parameters needed by the caller (*lastx, *eofp, *gotp, *prevp). Component 1: I created a new function in xfs_inode.c called xfs_iext_bno_to_ext(), which executes the binary search to find the target extent record. xfs_bmap_search_multi_extents() has been modified to call xfs_iext_bno_to_ext() rather than xfs_bmap_do_search_extents(). Component 2: The parameter setting functionality has been added to xfs_bmap_search_multi_extents(), eliminating the need for xfs_bmap_do_search_extents(). These changes make the removal of xfs_bmap_do_search_extents() trival once the CXFS changes are in place. They also allow us to maintain the current XFS interface, using the new search function introduced in mod xfs-linux:xfs-kern:207393a. SGI-PV: 928864 SGI-Modid: xfs-linux-melb:xfs-kern:207866a Signed-off-by: Mandy Kirkconnell Signed-off-by: Nathan Scott --- fs/xfs/xfs_bmap.c | 57 ++++++++++++++++++++++-------------------- fs/xfs/xfs_inode.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++++++++-- fs/xfs/xfs_inode.h | 1 + 3 files changed, 102 insertions(+), 29 deletions(-) (limited to 'fs/xfs/xfs_bmap.c') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index 64a02eaf1df..2d702e4a74a 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -3575,10 +3575,11 @@ xfs_bmap_do_search_extents( } /* - * Call xfs_bmap_do_search_extents() to search for the extent - * record containing block bno. If in multi-level in-core extent - * allocation mode, find and extract the target extent buffer, - * otherwise just use the direct extent list. + * Search the extent records for the entry containing block bno. + * If bno lies in a hole, point to the next entry. If bno lies + * past eof, *eofp will be set, and *prevp will contain the last + * entry (null if none). Else, *lastxp will be set to the index + * of the found entry; *gotp will contain the entry. */ xfs_bmbt_rec_t * /* pointer to found extent entry */ xfs_bmap_search_multi_extents( @@ -3589,36 +3590,38 @@ xfs_bmap_search_multi_extents( xfs_bmbt_irec_t *gotp, /* out: extent entry found */ xfs_bmbt_irec_t *prevp) /* out: previous extent entry found */ { - xfs_bmbt_rec_t *base; /* base of extent records */ xfs_bmbt_rec_t *ep; /* extent record pointer */ - xfs_ext_irec_t *erp = NULL; /* indirection array pointer */ xfs_extnum_t lastx; /* last extent index */ - xfs_extnum_t nextents; /* number of file extents */ /* - * For multi-level extent allocation mode, find the - * target extent list and pass only the contiguous - * list to xfs_bmap_do_search_extents. Convert lastx - * from a file extent index to an index within the - * target extent list. + * Initialize the extent entry structure to catch access to + * uninitialized br_startblock field. */ - if (ifp->if_flags & XFS_IFEXTIREC) { - int erp_idx = 0; - erp = xfs_iext_bno_to_irec(ifp, bno, &erp_idx); - base = erp->er_extbuf; - nextents = erp->er_extcount; - lastx = ifp->if_lastex - erp->er_extoff; - } else { - base = &ifp->if_u1.if_extents[0]; - nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); - lastx = ifp->if_lastex; + gotp->br_startoff = 0xffa5a5a5a5a5a5a5LL; + gotp->br_blockcount = 0xa55a5a5a5a5a5a5aLL; + gotp->br_state = XFS_EXT_INVALID; +#if XFS_BIG_BLKNOS + gotp->br_startblock = 0xffffa5a5a5a5a5a5LL; +#else + gotp->br_startblock = 0xffffa5a5; +#endif + prevp->br_startoff = NULLFILEOFF; + + ep = xfs_iext_bno_to_ext(ifp, bno, &lastx); + if (lastx > 0) { + xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx - 1), prevp); } - ep = xfs_bmap_do_search_extents(base, lastx, nextents, bno, - eofp, lastxp, gotp, prevp); - /* Convert lastx back to file-based index */ - if (ifp->if_flags & XFS_IFEXTIREC) { - *lastxp += erp->er_extoff; + if (lastx < (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))) { + xfs_bmbt_get_all(ep, gotp); + *eofp = 0; + } else { + if (lastx > 0) { + *gotp = *prevp; + } + *eofp = 1; + ep = NULL; } + *lastxp = lastx; return ep; } diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index 580fa075803..a16df2d435f 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -4265,12 +4265,81 @@ xfs_iext_destroy( ifp->if_bytes = 0; } +/* + * Return a pointer to the extent record for file system block bno. + */ +xfs_bmbt_rec_t * /* pointer to found extent record */ +xfs_iext_bno_to_ext( + xfs_ifork_t *ifp, /* inode fork pointer */ + xfs_fileoff_t bno, /* block number to search for */ + xfs_extnum_t *idxp) /* index of target extent */ +{ + xfs_bmbt_rec_t *base; /* pointer to first extent */ + xfs_filblks_t blockcount = 0; /* number of blocks in extent */ + xfs_bmbt_rec_t *ep = NULL; /* pointer to target extent */ + xfs_ext_irec_t *erp = NULL; /* indirection array pointer */ + int high; /* upper boundry in search */ + xfs_extnum_t idx = 0; /* index of target extent */ + int low; /* lower boundry in search */ + xfs_extnum_t nextents; /* number of file extents */ + xfs_fileoff_t startoff = 0; /* start offset of extent */ + + nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t); + if (nextents == 0) { + *idxp = 0; + return NULL; + } + low = 0; + if (ifp->if_flags & XFS_IFEXTIREC) { + /* Find target extent list */ + int erp_idx = 0; + erp = xfs_iext_bno_to_irec(ifp, bno, &erp_idx); + base = erp->er_extbuf; + high = erp->er_extcount - 1; + } else { + base = ifp->if_u1.if_extents; + high = nextents - 1; + } + /* Binary search extent records */ + while (low <= high) { + idx = (low + high) >> 1; + ep = base + idx; + startoff = xfs_bmbt_get_startoff(ep); + blockcount = xfs_bmbt_get_blockcount(ep); + if (bno < startoff) { + high = idx - 1; + } else if (bno >= startoff + blockcount) { + low = idx + 1; + } else { + /* Convert back to file-based extent index */ + if (ifp->if_flags & XFS_IFEXTIREC) { + idx += erp->er_extoff; + } + *idxp = idx; + return ep; + } + } + /* Convert back to file-based extent index */ + if (ifp->if_flags & XFS_IFEXTIREC) { + idx += erp->er_extoff; + } + if (bno >= startoff + blockcount) { + if (++idx == nextents) { + ep = NULL; + } else { + ep = xfs_iext_get_ext(ifp, idx); + } + } + *idxp = idx; + return ep; +} + /* * Return a pointer to the indirection array entry containing the * extent record for filesystem block bno. Store the index of the * target irec in *erp_idxp. */ -xfs_ext_irec_t * +xfs_ext_irec_t * /* pointer to found extent record */ xfs_iext_bno_to_irec( xfs_ifork_t *ifp, /* inode fork pointer */ xfs_fileoff_t bno, /* block number to search for */ @@ -4278,7 +4347,7 @@ xfs_iext_bno_to_irec( { xfs_ext_irec_t *erp = NULL; /* indirection array pointer */ xfs_ext_irec_t *erp_next; /* next indirection array entry */ - xfs_extnum_t erp_idx; /* indirection array index */ + int erp_idx; /* indirection array index */ int nlists; /* number of extent irec's (lists) */ int high; /* binary search upper limit */ int low; /* binary search lower limit */ diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h index 3c1df1d642f..006396764cb 100644 --- a/fs/xfs/xfs_inode.h +++ b/fs/xfs/xfs_inode.h @@ -478,6 +478,7 @@ void xfs_iext_indirect_to_direct(xfs_ifork_t *); void xfs_iext_direct_to_inline(xfs_ifork_t *, xfs_extnum_t); void xfs_iext_inline_to_direct(xfs_ifork_t *, int); void xfs_iext_destroy(xfs_ifork_t *); +xfs_bmbt_rec_t *xfs_iext_bno_to_ext(xfs_ifork_t *, xfs_fileoff_t, int *); xfs_ext_irec_t *xfs_iext_bno_to_irec(xfs_ifork_t *, xfs_fileoff_t, int *); xfs_ext_irec_t *xfs_iext_idx_to_irec(xfs_ifork_t *, xfs_extnum_t *, int *, int); void xfs_iext_irec_init(xfs_ifork_t *); -- cgit v1.2.3-70-g09d2